JP7159354B2 - Video signal processing method and apparatus using adaptive motion vector resolution - Google Patents

Description

The present invention relates to a video signal processing method and apparatus, and more particularly, to a video signal processing method and apparatus for encoding or decoding a video signal.

Compression coding refers to a series of signal processing techniques for transmitting digitized information through a communication line or storing it in a form suitable for a storage medium. Targets of compression encoding include audio, video, characters, and the like, and a technique of performing compression encoding especially for video is called video image compression. Compression coding of a video signal is performed by removing residual information in consideration of spatial correlation, temporal correlation, stochastic correlation, and the like. However, with the recent development of various media and data transmission media, there is a demand for a more efficient video signal processing method and apparatus.

SUMMARY OF THE INVENTION It is an object of the present invention to attempt to increase the coding efficiency of a video signal. Also, the present invention has an object to improve the signaling efficiency associated with the motion information set of the current block.

In order to solve the above problems, the present invention provides the following video signal processing apparatus and video signal processing method.

First, according to an embodiment of the present invention, in the video signal processing method, a motion vector predictor ( constructing a motion vector prediction (MVP) candidate list; obtaining a motion vector predictor for a current block based on said constructed MVP candidate list; between said current block motion vector and said motion vector predictor; obtaining a motion vector difference value indicating a difference between; modifying the motion vector difference value according to the resolution of the motion vector difference value of the current block, the resolution of the motion vector difference value being a resolution set is any one of the plurality of available resolutions included in the resolution set, and the configuration of the plurality of available resolutions included in the resolution set uses which method, out of the first method and the second method, for the current block. obtaining a motion vector for the current block based on the motion vector predictor and the modified motion vector difference value; and based on the obtained motion vector, said A video signal processing method is provided that includes the step of reconstructing a current block.

Also, according to an embodiment of the present invention, the video signal decoding apparatus includes a processor, wherein the processor performs motion compensation of a current block using any one of a first method and a second method. constructing a motion vector prediction (MVP) candidate list for, obtaining a motion vector predictor of a current block based on the constructed MVP candidate list, and generating the motion vector of the current block and the motion vector; obtaining a motion vector difference value indicating a difference between the predictor and the motion vector difference value, modifying the motion vector difference value based on the resolution of the motion vector difference value of the current block; The resolution is any one of the plurality of available resolutions included in the resolution set, and the configuration of the plurality of available resolutions included in the resolution set uses which of the first method and the second method. obtains a motion vector of the current block based on the motion vector predictor and the modified motion vector difference value, and uses the obtained motion vector as A video signal decoding apparatus is provided for reconstructing the current block based on.

The resolution of the motion vector difference value is a first resolution set and a second resolution, respectively, depending on which of the first method and the second method the MVP candidate list of the current block is constructed using. Obtained from any one of the sets, the second set of resolutions may include at least one other available resolution than the plurality of available resolutions included in the first set of resolutions.

When the MVP candidate list is constructed using the first method based on an affine model, the resolution of the motion vector difference values is obtained from the first resolution set, and the MVP candidate list is constructed using the affine model. When constructed using the second non-model-based method, the resolution of the motion vector difference values can be obtained from the second resolution set.

A first available resolution, which is the largest among the plurality of available resolutions included in the first resolution set, may be smaller than a second available resolution, which is the largest among the plurality of available resolutions included in the second resolution set.

The processor obtains an indicator indicating a resolution of a motion vector difference value of the current block among a plurality of available resolutions included in one of the first resolution set and the second resolution set, and The motion vector difference value can be modified based on the resolution indicated by the indicator. At this time, if the value of the indicator is the first value, the resolution indicated by the first value is determined by which method, out of the first method and the second method, the MVP candidate list uses. It can vary depending on how it is configured.

if the MVP candidate list is constructed using the first method, the first value indicates a first available resolution that is one of the available resolutions included in a first resolution set; and the MVP candidate list is constructed using the second method, the first value indicates a second available resolution that is one of the available resolutions included in the second resolution set, the first available resolution and the second The available resolutions may differ from each other.

When the first set of resolutions and the second set of resolutions both include a first available resolution, and the MVP candidate list is constructed using the second method, the first available resolution is the It can be indicated by a second value that is different from the first value.

The indicator may be represented by a variable length of bits, and the first value may be one of a plurality of values represented by the variable length of bits.

A third value different from the first value of the indicator is a value represented by the shortest length of bits among the plurality of values, and the MVP candidate list is constructed by the second method. , the third value indicates the smallest available resolution among a plurality of available resolution sets included in the second resolution set, and when the MVP candidate list is constructed by the first method, the third value is the A usable resolution other than the smallest usable resolution among the plurality of usable resolution sets included in the first resolution set can be indicated.

Also, according to an embodiment of the present invention, the video signal encoding apparatus includes a processor, wherein the processor generates a motion vector of the current block based on a position of a reference block referred to for motion compensation of the current block. and constructing a motion vector prediction (MVP) candidate list for motion compensation of the current block using any one of a first method and a second method, and obtaining the MVP candidate list. Obtaining a motion vector difference value based on a difference between any one of a plurality of candidates included in the list and the motion vector of the current block, and obtaining a motion vector difference value based on the resolution of the motion vector difference value of the current block to determine a motion vector difference value to be signaled, and wherein the resolution of the motion vector difference value is any one of a plurality of available resolutions included in a resolution set, and a plurality of available resolutions included in the resolution set. depends on which of the first method and the second method is used to construct the MVP candidate list of the current block, and a bitstream containing the signaled motion vector difference values An encoding apparatus is provided for generating.

The processor determines an indicator that indicates any one of a plurality of available resolutions included in any one of the first resolution set and the second resolution set, and determines the indicator and the signaled resolution. can generate a bitstream containing motion vector difference values. At this time, if the value of the indicator is the first value, the resolution indicated by the first value is determined by which method, out of the first method and the second method, the MVP candidate list uses. It can vary depending on how it is configured.

Further, according to an embodiment of the present invention, in a computer-readable recording medium storing a bitstream, the bitstream is modified based on the resolution of the motion vector difference value of the current block. including a modified motion vector difference value for a block, the resolution of the motion vector difference value being any one of a plurality of available resolutions included in a resolution set, and one of a plurality of available resolutions included in the resolution set. Configuration varies depending on which of the first method and the second method is used to construct a motion vector prediction (MVP) candidate list for motion compensation of the current block, bit A computer readable medium storing the stream is provided.

The bitstream may further include an indicator that indicates the resolution of the motion vector difference value of the current block among a plurality of available resolutions included in one of the first resolution set and the second resolution set. can be done. At this time, if the value of the indicator is the first value, the resolution indicated by the first value is determined by which method, out of the first method and the second method, the MVP candidate list uses. It can vary depending on how it is configured.

According to embodiments of the present invention, the coding efficiency of video signals can be enhanced. Also, according to an embodiment of the present invention, it is possible to increase the signaling efficiency for inter prediction of the current block.

1 is a schematic block diagram of a video signal encoding device according to one embodiment of the present invention; FIG. 1 is a schematic block diagram of a video signal decoding device according to one embodiment of the present invention; FIG. Fig. 3 shows an example in which a coding tree unit is split into coding units within a picture; FIG. 2 illustrates one embodiment of a method for signaling splitting of quadtrees and multi-type trees. FIG. 4 illustrates in more detail the intra prediction method according to an embodiment of the present invention; FIG. 4 illustrates in more detail the intra prediction method according to an embodiment of the present invention; 4 illustrates an inter-prediction method according to one embodiment of the present invention; FIG. 4 illustrates how the motion vector of the current block is signaled according to one embodiment of the present invention; FIG. 4 illustrates how a motion vector difference value of a current block is signaled according to an embodiment of the present invention; FIG. 4 illustrates how the resolution of the motion vector difference value of the current block is signaled according to an embodiment of the present invention; FIG. 4 is a diagram illustrating a method of signaling a resolution of a motion vector difference value of a current block according to another embodiment of the present invention; FIG. 4 is a diagram illustrating a method of signaling a resolution of a motion vector difference value of a current block according to another embodiment of the present invention; FIG. 11 illustrates an embodiment in which the configuration of available resolutions included in a resolution set changes; FIG. 10 is a diagram illustrating an embodiment of how the resolution of the motion vector difference value of the current block is signaled by the motion vector predictor of the current block. FIG. 10 is a diagram illustrating another embodiment of how the resolution of the motion vector difference value is signaled by the motion vector predictor of the current block; FIG. 4 illustrates how resolution of motion vector difference values is obtained based on a template matching method according to an embodiment of the present invention; FIG. 4 is a diagram illustrating how the resolution of motion vector difference values is obtained based on a two-sided matching method according to an embodiment of the present invention; FIG. 4 is a diagram illustrating a method of signaling a resolution of a motion vector difference value for each reference picture list of a current block according to an embodiment of the present invention; FIG. 4 is a diagram illustrating an embodiment of a method in which the resolution of motion vector difference values of the current block is signaled by the resolution of a picture; FIG. 4 illustrates how the resolution of motion vector difference values is signaled based on the size of the reference picture of the current block according to an embodiment of the present invention; 4 is a flow chart illustrating a method for obtaining a resolution of a motion vector difference value of a current block according to an embodiment of the present invention; 4 is a flow chart illustrating a method for obtaining a resolution of a motion vector difference value of a current block according to an embodiment of the present invention; 4 is a flow chart illustrating a method for obtaining a motion vector of a current block according to an embodiment of the present invention; 4 is a flow chart illustrating a method for obtaining a motion vector of a current block according to an embodiment of the present invention; FIG. 4 illustrates how a sign bit of a motion vector difference value of a current block is implicitly signaled according to an embodiment of the present invention; FIG. 4 illustrates how a sign bit of a motion vector difference value of a current block is implicitly signaled according to an embodiment of the present invention; FIG. 4 illustrates how a sign bit of a motion vector difference value of a current block is implicitly signaled according to an embodiment of the present invention; Figure 28 shows an example syntax for the embodiment of Figures 25-27; FIG. 4 is a diagram illustrating how the resolution of the motion vector difference value of the current block and the motion vector predictor are signaled according to one embodiment of the present invention; FIG. 4 illustrates a method of deriving a motion vector of a current block based on a plurality of motion vector difference values according to an embodiment of the present invention; FIG. 4 illustrates motion compensation based on an affine model according to one embodiment of the present invention; 1 illustrates one embodiment of a 4-parameter affine motion compensation method; 1 illustrates one embodiment of a 6-parameter affine motion compensation method; 1 illustrates an embodiment of a sub-block-based affine motion compensation method; Figure 4 illustrates an embodiment of a method for obtaining a control point motion vector set for prediction of the current block; Figure 4 illustrates an embodiment of a method for obtaining a control point motion vector set for prediction of the current block; Figure 4 illustrates an embodiment of a method for obtaining a control point motion vector set for prediction of the current block; Figure 4 illustrates an embodiment of a method for obtaining a control point motion vector set for prediction of the current block; FIG. 6 illustrates a method of obtaining a control point motion vector of a current block according to another embodiment of the present invention; 4 shows how control point motion vectors of the current block are obtained according to one embodiment of the present invention. 4 illustrates how control point motion vector difference values for the current block are signaled according to one embodiment of the present invention. FIG. 4 illustrates a method of obtaining a control point motion vector of a current block according to another embodiment of the present invention; FIG. FIG. 42 shows how a control point motion vector difference value is signaled when the control point motion vector of the current block is obtained according to the embodiment described with reference to FIG. FIG. 4 is a diagram illustrating how a control point motion vector difference value of a current block is signaled according to an embodiment of the present invention; FIG. 4 is a diagram illustrating how a motion vector is obtained using a differential predictor for a control point motion vector differential value of a current block according to an embodiment of the present invention; FIG. 4 illustrates how a control point motion vector of a current block is obtained using a differential predictor according to an embodiment of the present invention; FIG. 5 illustrates how a control point motion vector of a current block is obtained using a differential predictor according to another embodiment of the present invention; FIG. 4 illustrates how differential predictors are obtained according to one embodiment of the present invention; FIG. 4 illustrates how a difference vector predictor for the current block is determined according to one embodiment of the present invention; FIG. 4 is a diagram showing how a control point motion vector difference value of a current block is signaled; FIG. 4 is a diagram illustrating a method of deriving a control point motion vector of a current block according to another embodiment of the present invention; FIG. 4 illustrates how multiple control point motion vectors are derived for affine motion compensation of a current block according to an embodiment of the present invention; 4A and 4B illustrate various embodiments in which a current block is divided into a plurality of sub-blocks; FIG. 4 illustrates how a current block is divided according to an intra-prediction mode of the current block; FIG. FIG. 4 illustrates a method of dividing a current block into a plurality of sub-blocks based on sample values of reference samples of the current block according to an embodiment of the present invention; FIG. 4 illustrates how a primary sub-block group and a secondary sub-block group are determined according to an embodiment of the present invention; 4 illustrates how a primary group of sub-blocks and a secondary group of sub-blocks are predicted according to an embodiment of the present invention; 4 illustrates the order in which coding tree units are processed according to one embodiment of the present invention; 4 illustrates a bi-directional intra prediction method according to an embodiment of the present invention; 4 illustrates a method of predicting each of a plurality of sub-blocks divided from a current block according to an embodiment of the present invention;

The terms used in this specification have been selected as general terms that are currently widely used as much as possible while considering the functions in the present invention. It may vary depending on occurrences, etc. In addition, in certain cases, some terms are arbitrarily chosen by the applicant, and in such cases, their meanings are described in the detailed description of the invention. Therefore, it is clarified that the terms used in this specification should be construed based on the substantial meanings of the terms and the contents of the specification as a whole, not just the names of the terms.

In this specification, some terms are interpreted as follows. Coding is sometimes interpreted as encoding or coding. In this specification, a device that encodes a video signal to generate a video signal bitstream is referred to as an encoding device or encoder, and decodes the video signal bitstream to generate a video signal. is called a decoding device or decoder. Also, in this specification, a video signal processing device is used as a conceptual term including both an encoder and a decoder. Information is a term that includes values, parameters, coefficients, elements, etc., and may be interpreted differently depending on the case. The present invention is not limited to this. A 'unit' is used to refer to a basic unit of image processing or a specific position of a picture, and refers to an image area including at least one of a luminance (luma) component and a chrominance (chroma) component. Also, a "block" refers to an image area that includes a specific component of the luminance component and the color difference components (ie, Cb and Cr). However, terms such as 'unit', 'block', 'partition', and 'region' may be used interchangeably depending on the embodiment. Also, in this specification, a unit is used as a concept including all of a coding unit, a prediction unit, and a transform unit. A picture refers to a field or a frame, and the terms are used interchangeably depending on the embodiment.

FIG. 1 is a schematic block diagram of a video signal encoding device 100 according to one embodiment of the invention. Referring to FIG. 1, the encoding device 100 of this specification includes a transform unit 110, a quantization unit 115, an inverse quantization unit 120, an inverse transform unit 125, a filtering unit 130, a prediction unit 150, and an entropy coding unit 160. .

The transform unit 110 transforms the residual signal, which is the difference between the input video signal and the prediction signal generated by the prediction unit 150, to obtain a transform coefficient value. For example, Discrete Cosine Transform (DCT), Discrete Sine Transform (DST), Wavelet Transform, or the like is used. Discrete cosine transform and discrete sine transform are performed by dividing an input picture signal into blocks. In a transform, coding efficiency can vary depending on the distribution and characteristics of values within the transform domain. The quantization section 115 quantizes the values of the transform coefficients output from the transform section 110 .

In order to improve the coding efficiency, instead of coding the picture signal as it is, a picture is predicted using a pre-coded region through the prediction unit 150, and the predicted picture is a region between the original picture and the predicted picture. A method of adding residual values to obtain a reconstructed picture is used. To avoid mismatches between the encoder and decoder, the encoder should use information that is also available to the decoder when making predictions. To this end, the encoder performs a process of reconstructing the coded current block. The inverse quantization unit 120 inversely quantizes the transform coefficient values, and the inverse transform unit 125 restores the residual values using the inversely quantized transform coefficient values. Meanwhile, the filtering unit 130 performs a filtering operation to improve quality of the reconstructed picture and coding efficiency. For example, a deblocking filter, a Sample Adaptive Offset (SAO), an adaptive loop filter, etc. may be included. A filtered picture is stored in a decoded picture buffer (DPB) 156 to be output or used as a reference picture.

The predictor 150 includes an intra predictor 152 and an inter predictor 154 . An intra-predictor 152 performs intra-prediction within a current picture, and an inter-predictor 154 performs inter-prediction for predicting the current picture using a reference buffer stored in the decoded picture buffer 156 . The intra prediction unit 152 performs intra prediction from the reconstructed samples in the current picture and delivers intra coded information to the entropy coding unit 160 . The intra-encoding information includes at least one of an intra prediction mode, an MPM (Most Probable Mode) flag, and an MPM index. Intra-coded information includes information about reference samples. The inter prediction unit 154 includes a motion estimation unit 154a and a motion compensation unit 154b. The motion estimator 154a obtains a motion vector value of the current area by referring to a specific area of the reconstructed reference signal picture. The motion estimator 154 a transfers the motion information set (reference picture index, motion vector information) for the reference region to the entropy coding unit 160 . The motion compensator 154b performs motion compensation using the motion vector values received from the motion compensator 154a. The inter prediction unit 154 transfers inter coding information including the motion information set for the reference region to the entropy coding unit 160 .

According to a further embodiment, predictor 150 includes an intra block copy (BC) predictor (not shown). The intra BC predictor performs intra BC prediction from the reconstructed samples in the current picture and delivers intra BC coded information to the entropy coding unit 160 . The intra BC predictor obtains a block vector value indicating a reference region used for prediction of the current region by referring to a specific region within the current picture. The intra BC predictor performs intra BC prediction using the obtained block vector values. The intra BC predictor delivers the intra BC coded information to the entropy coder 160 . The intra BC predictor contains block vector information.

If the picture prediction described above is performed, the transform unit 110 transforms the residual values between the original picture and the predicted picture to obtain transform coefficient values. At this time, the transformation is performed on a specific block basis within a picture, and the size of the specific block is variable within a preset range. The quantization unit 115 quantizes the values of the transform coefficients generated by the transform unit 110 and transfers the quantized values to the entropy coding unit 160 .

The entropy coding unit 160 entropy-codes the quantized transform coefficients, intra-encoded information, inter-encoded information, and the like to generate a video signal bitstream. The entropy coding unit 160 uses a variable length coding (VLC) scheme, an arithmetic coding scheme, and the like. A variable length coding (VLC) scheme converts incoming symbols into a sequence of codewords, but the length of the codewords is variable. For example, frequently occurring symbols are represented by short codewords, and infrequently occurring symbols are represented by long codewords. A context-based adaptive variable length coding (CAVLC) scheme is used as the variable length coding scheme. Arithmetic coding converts consecutive data symbols into a single prime, but arithmetic coding obtains the optimal prime bits needed to represent each symbol. A Context-based Adaptive Binary Arithmetic Coding (CABAC) scheme is used as the arithmetic coding.

The generated bitstream is encapsulated using a Network Abstraction Layer (NAL) unit as a basic unit. A NAL unit includes an integer number of coding tree units that are encoded. In order to decode a bitstream in a video decoder, the bitstream should be first separated into NAL units and then each separated NAL unit should be decoded. Information necessary for decoding a video signal bitstream includes a picture parameter set (PPS), a sequence parameter set (SPS), a video parameter set (VPS), and the like. is transmitted via a higher level set of RBSP (Raw Byte Sequence Payload) such as .

Meanwhile, the block diagram of FIG. 1 shows an encoding apparatus 100 according to an embodiment of the present invention, and separated blocks logically distinguish elements of the encoding apparatus 100. As shown in FIG. Thus, the elements of encoding apparatus 100 described above may be mounted on a single chip or multiple chips depending on the design of the device. According to one embodiment, the operations of each element of encoding apparatus 100 described above are performed by a processor (not shown).

FIG. 2 is a schematic block diagram 200 of a video signal decoding apparatus according to an embodiment of the invention. Referring to FIG. 2 , the decoding device 200 of this specification includes an entropy decoding unit 210 , an inverse quantization unit 220 , an inverse transform unit 225 , a filtering unit 230 and a prediction unit 250 .

The entropy decoding unit 210 entropy-decodes the video signal bitstream to extract transform coefficients, intra-encoding information, inter-encoding information, etc. for each region. An inverse quantizer 220 inversely quantizes the entropy-decoded transform coefficients, and an inverse transform unit 225 restores residual values using the inversely quantized transform coefficients. The video signal processing apparatus 200 sums the residual values obtained from the inverse transform unit 225 with the predictors obtained from the prediction unit 250 to restore the original pixel values.

Meanwhile, the filtering unit 230 performs filtering on pictures to improve picture quality. This includes a deblocking filter to reduce block distortion phenomena and/or an adaptive loop filter to remove distortion of the entire picture. The filtered picture is either output or stored in a decoded picture buffer (DPB) 256 for use as a reference picture for the next picture.

The predictor 250 includes an intra predictor 252 and an inter predictor 254 . The prediction unit 250 generates a prediction picture using the coding type, transform coefficients for each region, intra/inter coding information, etc. decoded by the entropy decoding unit 210 described above. A decoded region of a current picture or another picture containing the current block can be used to reconstruct the current block on which decoding is performed. A picture (or tile/slice) that uses only the current picture for reconstruction, i.e. performs only intra prediction or intra BC prediction, is called intra picture or I picture (or tile/slice), intra prediction, inter prediction and intra BC prediction. is called an inter-picture (or tile/slice). Predictive picture or P A picture (or tile/slice) is called a picture (or tile/slice), and a picture (or tile/slice) that uses up to two motion vectors and a reference picture index is called a bi-predictive picture or a B picture (or tile/slice). It says. In other words, P pictures (or tiles/slices) use at most one set of motion information to predict each block, and B pictures (or tiles/slices) use at most two motion information sets to predict each block. Use the motion information set. Here, the motion information set includes one or more motion vectors and one reference picture index.

The intra prediction unit 252 generates a prediction block using the intra coded information and the reconstructed samples in the current picture. As described above, the intra-encoding information includes at least one of an intra prediction mode, an MPM (MOST Probable Mode) flag, and an MPM index. The intra prediction unit 252 predicts sample values of the current block using the reconstructed samples positioned to the left and/or above the current block as reference samples. In this disclosure, reconstructed samples, reference samples, and current block samples refer to pixels. Also, a sample value indicates a pixel value.

In one embodiment, the reference samples are samples contained in neighboring blocks of the current block. For example, the reference samples are the samples adjacent to the left boundary and/or the samples adjacent to the upper boundary of the current block. Also, the reference samples are samples located on a line within a preset distance from the left boundary of the current block and/or samples within a preset distance from the upper boundary of the current block among samples of neighboring blocks of the current block. It is a sample located on the line. At this time, the neighboring blocks of the current block are the left (L) block, the upper (A) block, the Below Left (BL) block, the Above Right (AR) block, or the upper left block adjacent to the current block. (Above Left, AL) contains at least one of the blocks.

The inter prediction unit 254 generates a prediction block using reference pictures and inter coding information stored in the decoded picture buffer 256 . The inter-coded information includes the motion information set (reference picture index, motion vector, etc.) of the current block relative to the reference block. Inter prediction includes L0 prediction, L1 prediction, and bi-prediction. L0 prediction is prediction using one reference picture included in the L0 picture list, and L1 prediction is prediction using one reference picture included in the L1 picture list. For that, a set of motion information (eg, motion vectors and reference picture indices) is required. The bi-prediction method uses a maximum of two reference regions, and these two reference regions may exist in the same reference picture or in different pictures. That is, the bi-prediction scheme uses a maximum of two sets of motion information (e.g., motion vector and reference picture index), but two motion vectors may correspond to the same reference picture index, or may correspond to different reference picture indices. You can respond. At this time, the reference picture is temporally displayed (or output) either before or after the current picture.

The inter predictor 254 obtains a current reference block using motion vectors and reference picture indices. The reference block resides in a reference picture corresponding to a reference picture index. Also, a sample value of the block specified by the motion vector or an interpolation value thereof is used as a predictor of the current block. For motion prediction with sub-pel pixel accuracy, for example, an 8-tab interpolation filter is used for luminance signals and a 4-tab interpolation filter for chrominance signals. However, the interpolation filter for sub-pel unit motion prediction is not limited to this. Thus, the inter predictor 254 performs motion compensation to predict the texture of the current unit from the previously reconstructed picture. At this time, the inter predictor uses the motion information set.

According to additional embodiments, predictor 250 can include an intra BC predictor (not shown). The intra BC predictor performs intra BC prediction from the reconstructed samples in the current picture and delivers intra BC coded information to the entropy coder 160 . The intra BC predictor obtains a block vector value of the current region that indicates a specific region within the current picture. The intra BC predictor can perform intra BC prediction using the obtained block vector values. The intra BC predictor delivers the intra BC coded information to the entropy coder 160 . Intra BC encoded information may include block vector information.

A reconstructed video picture is generated by adding the predictors output from the intra predictor 252 or the inter predictor 254 and the residual values output from the inverse transform unit 225 . That is, the video signal decoding apparatus 200 restores the current block using the prediction block generated by the prediction unit 250 and the residual obtained from the inverse transform unit 225 .

Meanwhile, the block diagram of FIG. 2 illustrates a decoding apparatus 200 according to an embodiment of the present invention, and separated blocks logically distinguish elements of the decoding apparatus 200 . Thus, the elements of decoding apparatus 200 described above may be mounted on a single chip or multiple chips depending on the design of the device. According to one embodiment, the operations of each element of the decoding device 200 described above are performed by a processor (not shown).

FIG. 3 shows an example where a Coding Tree Unit (CTU) is divided into Coding Units (CUs) within a picture. During the coding process of a video signal, pictures are divided into a sequence of coding tree units (CTUs). A coding tree unit consists of an NXN block of luma samples and two corresponding blocks of chrominance samples. A coding tree unit is divided into multiple coding units. A coding tree unit may become a leaf node without being split. In this case, the coding tree unit itself can be a coding unit. A coding unit refers to a basic unit for processing a picture in the above video signal processing process, ie, intra/inter prediction, transform, quantization, and/or entropy coding. Within one picture, the size and pattern of coding units are not constant. The coding units have a square or rectangular pattern. A rectangular coding unit (or rectangular block) includes a vertical coding unit (or vertical block) and a horizontal coding unit (or horizontal block). As used herein, a vertical block is a block whose height is greater than its width, and a horizontal block is a block whose width is greater than its height. Also, in this specification, a non-square block refers to a rectangular block, but the invention is not so limited.

Referring to FIG. 3, the coding tree unit is first divided into a Quad Tree (QT) structure. That is, in the quadtree structure, one node with a size of 2Nx2N is divided into four nodes with a size of NxN. A quadtree is also referred to herein as a quaternary tree. Quadtree splitting is recursive and not all nodes need to be split to the same depth.

Meanwhile, the leaf nodes of the quadtree described above are further divided into a multi-type tree (MTT) structure. According to an embodiment of the present invention, a multi-type tree structure divides one node into a horizontal or vertical split binary or ternary tree structure. That is, there are four partitioning structures in the multi-type tree structure: vertical binary partitioning, horizontal binary partitioning, vertical ternary partitioning, and horizontal ternary partitioning. According to an embodiment of the present invention, in each tree structure, the width and height of a node both have powers of two. For example, in a binary tree (BT) structure, a node of size 2Nx2N is split into two Nx2N nodes by vertical binary splitting, and split into two 2NxN nodes by horizontal binary splitting. Also, in a ternary tree (TT) structure, a 2N×2N size node is divided into (N/2)×2N, N×2N and (N/2)×2N nodes by vertical ternary partitioning. , is divided into 2N×(N/2), 2N×N and 2N×(N/2) nodes by horizontal ternary partitioning. Such multi-type tree splitting is done recursively.

A leaf node of a multi-type tree can be a coding unit. If no division for a coding unit is indicated or the coding unit is not larger than the maximum transform length, the corresponding coding unit is used as a prediction and transform unit without further division. Meanwhile, in the quadtree and multi-type tree described above, at least one of the following parameters is predefined or transmitted via a higher level set of RBSPs such as PPS, SPS, VPS, and so on. 1) CTU size: the size of the root node of the quadtree, 2) MinQtSize: the size of the smallest QT leaf node allowed, 3) MaxBtSize: the largest BT allowed. Root node size, 4) Maximum TT size (MaxTtSize): Maximum allowed TT root node size, 5) Maximum MTT depth (MaxMttDepth): Maximum allowed depth of MTT splits from QT leaf nodes, 6) Minimum BT size (MinBtSize): size of the smallest BT leaf node allowed; 7) Minimum TT size: size of the smallest TT leaf node allowed.

FIG. 4 illustrates one embodiment of a method for signaling splitting of quadtrees and multi-type trees. A preset flag can be used to signal the splitting of the quadtree and multi-type tree described above. Referring to FIG. 4, a flag 'qt_split_flag' indicating whether to split a quadtree node, a flag 'mtt_split_flag' indicating whether to split a multi-type tree node, and a split direction of a multi-type tree node. At least one of the flag 'mtt_split_vertical_flag' indicating the split form of the multi-type tree node and the flag 'mtt_split_binary_flag' indicating the split form of the multi-type tree node can be used.

According to embodiments of the present invention, the coding tree unit is the root node of the quadtree and can be pre-divided into the quadtree structure. In the quadtree structure, 'qt_split_flag' is signaled for each node 'QT_node'. If the value of 'qt_split_flag' is 1, the corresponding node is split into 4 square nodes, and if the value of 'qt_split_flag' is 0, the corresponding node is a quadtree leaf node 'QT_leaf_node'.

Each quadtree leaf node 'QT_leaf_node' can be further divided into a multi-type tree structure. In the multi-type tree structure, 'mtt_split_flag' is signaled for each node 'MTT_node'. If the value of 'mtt_split_flag' is 1, the corresponding node is split into multiple rectangular nodes, and if the value of 'mtt_split_flag' is 0, the corresponding node becomes a leaf node 'MTT_leaf_node' of the multi-type tree. 'mtt_split_vertical_flag' and 'mtt_split_binary_flag' for node 'MTT_node' additionally signaling when multi-type tree node 'MTT_node' is split into multiple rectangular nodes (i.e. 'mtt_split_flag' value is 1) can. If the value of 'mtt_split_vertical_flag' is 1, vertical splitting of the node 'MTT_node' is indicated, and if the value of 'mtt_split_vertical_flag' is 0, horizontal splitting of the node 'MTT_node' is indicated. Also, if the value of 'mtt_split_binary_flag' is 1, the node 'MTT_node' is split into two rectangular nodes, and if the value of 'mtt_split_binary_flag' is 0, the node 'MTT_node' is split into 3 rectangular nodes.

Picture prediction (motion compensation) for coding is performed on coding units that cannot be further divided (ie, leaf nodes of the coding unit tree). A basic unit for such prediction is hereinafter referred to as a prediction unit or a prediction block.

Hereinafter, the term "unit" used in this specification is used as a substitute term for the prediction unit, which is a basic unit for prediction. However, the present invention is not limited to this, and in a broader sense, it is understood as a concept including the coding unit.

5 and 6 are diagrams illustrating the intra prediction method in more detail according to an embodiment of the present invention. As described above, the intra predictor predicts the sample values of the current block using the reconstructed samples positioned to the left and/or above the current block as reference samples.

First, FIG. 5 shows an example of reference samples used to predict the current block in intra-prediction mode. According to one embodiment, the reference samples are the samples adjacent to the left boundary and/or the samples adjacent to the upper boundary of the current block. As shown in FIG. 5, if the size of the current block is W×H and the samples of a single reference line (line) adjacent to the current block are used for intra prediction, then the current block is positioned to the left and/or above the current block. A maximum of 2W+2H+1 surrounding samples are used to set the reference sample.

According to additional embodiments of the present invention, samples on multiple reference lines can be used for intra-prediction of the current block. The plurality of reference lines may consist of n lines positioned within a preset distance from the boundary of the current block. In this case, separate reference line information indicating at least one reference line used for intra prediction of the current block can be signaled. Specifically, the reference line information may include an index designating one of a plurality of reference lines.

Also, if at least some of the samples used as reference samples have not been reconstructed, the intra predictor obtains reference samples by performing a reference sample padding process. In addition, the intra predictor performs a reference sample filtering process to reduce intra prediction errors. That is, the filtered reference samples are obtained by filtering the reference samples obtained by the surrounding samples and/or the reference sample padding process. The intra predictor predicts samples of the current block using unfiltered reference samples or filtered reference samples. In the present disclosure, peripheral samples include samples on at least one reference line. For example, surrounding samples may include adjacent samples on lines adjacent to the boundary of the current block.

Next, FIG. 6 illustrates one embodiment of prediction modes used for intra prediction. For intra prediction, intra prediction mode information can be signaled that indicates the intra prediction direction. The intra-prediction mode information indicates any one of a plurality of intra-prediction modes forming an intra-prediction mode set. If the current block is an intra-predicted block, the decoder receives intra-prediction mode information of the current block from the bitstream. An intra prediction unit of the decoder performs intra prediction for the current block based on the extracted intra prediction mode information.

According to an embodiment of the present invention, an intra-prediction mode set includes all intra-prediction modes used for intra-prediction (eg, a total of 67 intra-prediction modes). More specifically, the intra-prediction mode set includes planar mode, DC mode, and multiple (eg, 65) angular modes (ie, directional modes). In some embodiments, an intra-prediction mode set may consist of a subset of all intra-prediction modes. Each intra-prediction mode is indicated via a preset index (ie, intra-prediction mode index). For example, as shown in FIG. 6, an intra prediction mode index of 0 indicates a planar mode, and an intra prediction mode index of 1 indicates a DC mode. In addition, intra prediction mode indexes 2 to 66 respectively indicate different angular modes. At this time, the intra prediction mode index 2 indicates a horizontal diagonal (HDIA) mode, the intra prediction mode index 18 indicates a horizontal (HOR) mode, and the intra prediction mode index 34 indicates a diagonal (Diagonal) mode. , DIA) mode, an intra prediction mode index 50 indicates a vertical (VER) mode, and an intra prediction mode index 66 indicates a vertical diagonal (VDIA) mode.

Hereinafter, an inter prediction method according to an embodiment of the present invention will be described with reference to FIG. In the present disclosure, the inter-prediction method may include a general inter-prediction method optimized for translational motion and an affine model-based inter-prediction method, which will be described later with reference to FIGS. Also, the motion vector may include at least one of a general motion vector for motion compensation according to a general inter-prediction method and a control point motion vector for affine motion compensation.

FIG. 7 illustrates an inter-prediction method according to one embodiment of the present invention. As described above, the decoder can refer to the reconstructed samples of other decoded pictures to predict the current block. Referring to FIG. 7, the decoder obtains reference block 702 in reference picture 720 based on the motion information set of current block 701 . At this time, the motion information set can include reference picture indices and motion vectors 703 . The reference picture index indicates a reference picture 720 including a reference block for inter prediction of the current block in the reference picture list. According to one embodiment, the reference picture list may include at least one of the aforementioned L0 picture list or L1 picture list. Motion vector 703 indicates the offset between the coordinates of current block 701 in current picture 710 and the coordinates of reference block 702 in reference picture 720 . The decoder obtains a predictor of the current block 701 based on the sample values of the reference block 702 and reconstructs the current block 701 using the predictor.

Specifically, the encoder can acquire the above-described reference block by searching for a block similar to the current block in a picture whose reconstruction order is earlier. For example, the encoder can search for a reference block that minimizes the sum of differences between the current block and sample values within a preset search area. At this time, at least one of SAD (Sum Of Absolute Difference) or SATD (Sum Of Hadamard Transformed Difference) may be used to measure the similarity between samples of the current block and the reference block. Here, the SAD may be a sum of absolute values of differences between sample values included in two blocks. Also, the SATD may be a sum of absolute values of Hadamard transform coefficients obtained by Hadamard transforming the difference between sample values included in two blocks.

Meanwhile, the current block can also be predicted using one or more reference regions. As described above, the current block can be inter-predicted by a bi-prediction scheme using two or more reference regions. According to one embodiment, the decoder can obtain two reference blocks based on two motion information sets of the current block. Also, the decoder can obtain a first predictor and a second predictor of the current block based on the obtained sample values of each of the two reference blocks. Also, the decoder can reconstruct the current block using the first predictor and the second predictor. For example, the decoder can reconstruct the current block based on sample-by-sample averages of the first predictor and the second predictor.

As described above, one or more motion information sets can be signaled for motion compensation of the current block. At this time, the similarity between motion information sets for motion compensation of each of a plurality of blocks can be used. For example, the motion information set used for prediction of the current block can be derived from the motion information set used for prediction of one of the previously reconstructed samples. Through this, the encoder and decoder can reduce signaling overhead. Hereinafter, various embodiments in which the motion information set of the current block is signaled will be described.

FIG. 8 is a diagram illustrating how the motion vector of the current block is signaled according to one embodiment of the invention. According to one embodiment of the present invention, the motion vector of the current block can be derived from a motion vector predictor (MVP) of the current block. According to one embodiment, a motion vector predictor that is referenced to derive the motion vector of the current block can be obtained using a motion vector predictor (MVP) candidate list. The MVP candidate list can include a preset number of MVP candidates (Candidate 1, Candidate 2, ..., Candidate N).

According to one embodiment, the MVP candidate list may include at least one of spatial candidates or temporal candidates. A spatial candidate may be a motion information set used for prediction of neighboring blocks within a certain range from the current block in the current picture. Spatial candidates can be constructed based on available neighboring blocks of the current block. Also, the temporal candidates may be motion information sets used for prediction of blocks in the current picture and other pictures. For example, temporal candidates can be constructed based on a particular block corresponding to the location of the current block within a particular reference picture. At this time, the position of the specific block indicates the position of the top-left sample of the specific block within the reference picture. According to additional embodiments, the MVP candidate list may contain zero motion vectors. According to additional embodiments, a rounding process for MVP candidates included in the current block's MVP candidate list may be performed. At this time, the resolution of the motion vector difference value of the current block, which will be described later, can be used. For example, each MVP candidate of the current block can be rounded based on the resolution of the motion vector difference value of the current block.

In this disclosure, the MVP candidate list is an advanced temporal motion vector prediction (ATMVP) list, a merge candidate list for merged inter prediction, a control point motion vector candidate list for affine motion compensation, It can include a subblock-based temporal motion vector prediction (STMVP) list for subblock-based motion compensation, and combinations thereof.

According to one embodiment, encoder 810 and decoder 820 can construct an MVP candidate list for motion compensation of the current block. For example, among samples reconstructed prior to the current block, there may be candidates corresponding to samples that may have been predicted based on the same or similar motion information set as the motion information set of the current block. . The encoder 810 and decoder 820 may construct an MVP candidate list of the current block based on the corresponding candidate blocks. At this time, the encoder 810 and the decoder 820 can construct the MVP candidate list according to predefined rules between the encoder 810 and the decoder 820 . That is, the MVP candidate lists configured in each of the encoder 810 and decoder 820 may be the same.

Also, the predefined rule may vary according to the prediction mode of the current block. For example, if the prediction mode of the current block is the affine model-based affine prediction mode, the encoder and decoder can construct the MVP candidate list of the current block using the first method based on the affine model. A first method may be a method of obtaining a control point motion vector candidate list. This will be described in detail with reference to FIGS. 31 to 52. FIG. On the other hand, if the prediction mode of the current block is the general inter prediction mode that is not based on the affine model, the encoder and decoder can construct the MVP candidate list of the current block using the second method that is not based on the affine model. At this time, the first method and the second method may be different methods.

The decoder 820 may derive a motion vector of the current block based on any one of at least one MVP candidate included in the MVP candidate list of the current block. For example, the encoder 810 can signal an MVP index that indicates a motion vector predictor that is referenced to derive the motion vector of the current block. Decoder 820 can obtain a motion vector predictor for the current block based on the signaled MVP index. The decoder 820 can derive the motion vector of the current block using the motion vector predictor. According to one embodiment, the decoder 820 can use the motion vector predictor obtained from the MVP candidate list for the motion vector of the current block without a separate motion vector difference value. A decoder 820 can reconstruct the current block based on the motion vector of the current block. An inter-prediction mode in which a motion vector predictor obtained from the MVP candidate list is used for the motion vector of the current block without a separate motion vector difference value may be referred to as a merge mode.

According to another embodiment, the decoder 820 can obtain a separate motion vector difference for the motion vector of the current block. The decoder 820 may obtain a motion vector of the current block by summing the motion vector predictor obtained from the MVP candidate list and the motion vector difference value of the current block. In this case, the encoder 810 can signal a motion vector difference value (MV difference) that indicates the difference between the motion vector of the current block and the motion vector predictor. A method of signaling the motion vector difference value will be described in detail with reference to FIG. The decoder 820 can obtain the motion vector of the current block based on the motion vector difference (MV difference). A decoder 820 can reconstruct the current block based on the motion vector of the current block.

Additionally, a reference picture index for motion compensation of the current block can be signaled. If the prediction mode of the current block is the encoder 810 can signal a reference picture index that indicates the reference picture that contains the reference block. The decoder 820 can obtain a POC of a reference picture referenced to reconstruct the current block based on the signaled reference picture index. At this time, the POC of the reference picture may differ from the POC of the reference picture corresponding to the MVP referred to for deriving the motion vector of the current block. In this case, the decoder 820 can perform motion vector scaling. That is, decoder 820 can scale MVP to obtain MVP'. At this time, the motion vector scaling may be performed based on the POC of the current picture, the POC of the signaled reference picture of the current block, and the POC of the reference picture corresponding to the MVP. Decoder 820 can also use MVP' as a motion vector predictor for the current block.

As described above, the motion vector of the current block can be obtained by summing the motion vector predictor of the current block and the motion vector difference value. At this time, the motion vector difference value can be signaled from the encoder. An encoder can encode the motion vector differential values to generate and signal information indicative of the motion vector differential values. The following describes how motion vector difference values are signaled according to one embodiment of the present invention.

FIG. 9 is a diagram illustrating how the motion vector difference value of the current block is signaled according to one embodiment of the present invention. According to one embodiment, the information indicating the motion vector difference value may include at least one of absolute value information of the motion vector difference value and sign information of the motion vector difference value. The absolute value and sign of the motion vector difference value can be encoded separately.

According to one embodiment, the absolute value of the motion vector difference value may not be signaled in the value itself. The encoder can reduce the size of the value signaled using at least one flag that indicates the properties of the absolute value of the motion vector difference value. The decoder can derive the absolute value of the motion vector difference value using at least one flag from the signaled value.

For example, the at least one flag can include a first flag indicating whether the absolute value of the motion vector difference value is greater than N. At this time, N may be an integer. If the size of the absolute value of the motion vector difference value is greater than N, the (absolute value of the motion vector difference value−N) value can be signaled with the first flag activated. At this time, the activated flag may indicate that the size of the absolute value of the motion vector difference value is greater than N. The decoder can obtain the absolute value of the motion vector difference value based on the activated first flag and the signaled value.

Referring to FIG. 9, a second flag (abs_mvd_greater0_flag) indicating whether the absolute value of the motion vector difference value is greater than '0' can be signaled. When the second flag (abs_mvd_greater0_flag[]) indicates that the absolute value of the motion vector difference value is not greater than '0', the absolute value of the motion vector difference value may be '0'. Also, if the second flag (abs_mvd_greater0_flag) indicates that the absolute value of the motion vector difference value is greater than '0', the decoder obtains the absolute value of the motion vector difference value using other information about the motion vector difference value. be able to.

According to one embodiment, a third flag (abs_mvd_greater1_flag) indicating whether the absolute value of the motion vector difference value is greater than '1' can be signaled. If the third flag (abs_mvd_greater1_flag) indicates that the absolute value of the motion vector difference value is not greater than '1', the decoder can determine that the absolute value of the motion vector difference value is '1'.

Conversely, if the third flag (abs_mvd_greater1_flag) indicates that the absolute value of the motion vector difference value is greater than '1', the decoder uses further information about the motion vector difference value to determine the absolute value of the motion vector difference value. can be obtained. For example, (abs_mvd_minus2) (abs_mvd_minus2) (absolute value of motion vector difference value minus 2) can be signaled. This is because when the absolute value of the motion vector difference value is greater than '1', the absolute value of the motion vector difference value can be 2 or more.

As described above, the absolute value of the motion vector difference value of the current block can be transformed into at least one flag. For example, the transformed absolute value of the motion vector differential value can be indicated by the size of the motion vector differential value (absolute value of motion vector differential value−N). According to one embodiment, the transformed absolute value of the motion vector difference value can be signaled through at least one bit. At this time, the number of bits signaled to indicate the transformed absolute value of the motion vector difference value may be variable. The encoder can encode the transformed absolute value of the motion vector difference value using a variable length binarization method. For example, the encoder uses at least one of truncated unary binarization, unary binarization, truncated rice, or exp-Golomb binarization. one can be used.

Also, the sign of the motion vector difference value can be signaled through a sign flag (mvd_sign_flag). On the other hand, the sign of the motion vector difference value can also be implicitly signaled by sign-bit-hiding. This will be described with reference to FIGS. 23 through 28. FIG.

Meanwhile, the above motion vector difference value of the current block can be signaled in a specific resolution unit. In the present disclosure, the resolution of the motion vector difference value may indicate the units in which the motion vector difference value is signaled. That is, in this disclosure, the resolution, excluding the picture resolution, can indicate the precision or granularity with which the motion vector difference values are signaled. The resolution of motion vector difference values can be expressed in units of samples or pixels. For example, the resolution of the motion vector difference value can be expressed using a sample unit such as 1/4 (quarter), 1/2 (half), 1 (integer), 2, or 4 sample units. In addition, as the resolution of the motion vector difference value of the current block is smaller, the accuracy of the motion vector difference value of the current block can be increased.

According to one embodiment of the present invention, motion vector difference values can be signaled based on various resolutions. According to one embodiment, the absolute value or the transformed absolute value of the motion vector difference value can be signaled in integer sample units. Alternatively, the absolute value of the motion vector difference value can be signaled in 1/2-subpel units. That is, the resolution of the motion vector difference value can be set differently depending on the situation. An encoder and decoder according to an embodiment of the present invention can efficiently signal the motion vector difference value of the current block by appropriately utilizing various resolutions for the motion vector difference value.

According to one embodiment, the resolution of motion vector difference values can be set to different values for at least one of blocks, coding units, slices, or tiles. For example, the first resolution of the motion vector difference value of the first block may be 1/4 sample unit. In this case, '64', which is a value obtained by dividing the absolute value '16' of the motion vector difference value by the first resolution, can be signaled. Also, the second resolution of the motion vector difference value of the second block may be an integer sample unit. In this case, '16', which is a value obtained by dividing the absolute value '16' of the second motion vector difference value by the second resolution, can be signaled. In this way, even when the absolute values of the motion vector difference values are the same, different values can be signaled depending on the resolution. At this time, when the value obtained by dividing the absolute value of the motion vector difference value by the resolution includes a decimal point or less, a rounding function can be applied to the corresponding value.

The encoder can signal information indicative of the motion vector difference values based on the resolution of the motion vector difference values. The decoder can obtain modified motion vector difference values from the signaled motion vector difference values. A decoder can modify the motion vector difference values based on the resolution of the resolution difference values. The relationship between the signaled motion vector difference value (valuePerResoultion) of the current block and the modified motion vector difference value (valueDetermined) is as follows [Equation 1]. Hereinafter, in the present disclosure, motion vector difference values refer to modified motion vector difference values (valueDetermined) unless otherwise specified. Also, the signaled motion vector difference value indicates the value before being modified by the resolution.

In [Equation 1], resolution indicates the resolution of the motion vector difference value of the current block. That is, the decoder can obtain the modified motion vector difference value by multiplying the signaled motion vector difference value of the current block by the resolution. Next, the decoder can obtain the motion vector of the current block based on the modified motion vector difference value. Also, the decoder can reconstruct the current block based on the motion vector of the current block.

If a relatively small value is used for the resolution of the motion vector difference values of the current block (ie, if the accuracy is high), it may be advantageous to indicate the motion vector difference values of the current block more precisely. However, in this case, the value itself to be signaled is large, and the signaling overhead for the motion vector difference value of the current block may increase. On the contrary, when a relatively large value is used for the resolution of the motion vector difference value of the current block (i.e., when the accuracy is low), the size of the signaled value is reduced to reduce the signaling overhead for the motion vector difference value. can be reduced. That is, when the resolution of the motion vector difference value is high, the motion vector difference value of the current block can be signaled through fewer bits than when the resolution of the motion vector difference value of the current block is low. However, in this case, it may be difficult to precisely express the motion vector difference value of the current block.

This allows the encoder and decoder to select an advantageous resolution for signaling the motion vector difference value among multiple resolutions depending on the situation. For example, the encoder can signal the selected resolution based on context. Also, the decoder can obtain the motion vector difference value of the current block based on the signaled resolution. Hereinafter, a method of signaling the resolution of the motion vector difference value of the current block according to an embodiment of the present invention will be described. According to an embodiment of the present invention, the resolution of the motion vector difference value of the current block may be any one of a plurality of available resolutions included in the resolution set. Here, a plurality of available resolutions can indicate resolutions that can be used in a particular situation. Also, the types and number of available resolutions included in the resolution set may vary according to circumstances.

FIG. 10 is a diagram illustrating how the resolution of the motion vector difference value of the current block is signaled according to an embodiment of the present invention. According to an embodiment, a resolution indicator can be signaled that indicates the resolution of the motion vector difference value of the current block among a plurality of available resolutions. The encoder can signal the resolution indicator along with information indicating motion vector difference values. A decoder can obtain the resolution of the motion vector difference value of the current block based on a resolution indicator that indicates one of a plurality of available resolutions included in the resolution set of the current block. A decoder can obtain the motion vector difference value of the current block based on the resolution of the obtained motion vector difference value.

According to one embodiment, the resolution indicator can be represented by bits of variable length. For example, the resolution indicator may indicate one of resolution indices that indicate each of a plurality of available resolutions in the resolution set of the current block. At this time, the resolution index can be represented by variable length bits having a preset maximum length. For example, the resolution indicator can include two or more flags each represented by one bit. The preset maximum length may be a length determined by the number of available resolutions. For example, if the available resolution set includes 3 available resolutions, the preset maximum length may be '2'. In this disclosure, the resolution index may also be referred to as the value of the resolution indicator. Referring to FIG. 10, the value of the resolution indicator can be any one of 0, 10, or 11.

In the embodiment of FIG. 10, the resolution set for motion compensation of the current block may include available resolutions of 1/4, 1, and 4 samples. At this time, the resolution of the motion vector difference value of the current block may be any one of available resolutions (1/4, 1, 4) shown in FIG. According to one embodiment, of the multiple available resolutions, the smallest available resolution (ie, the resolution with the highest accuracy) can be signaled by an indicator value that uses the shortest length bits. For example, among available resolutions of 1/4, 1, and 4 sample units, the smallest available resolution of 1/4 sample unit is indicated by '0', which is an indicator value that uses the shortest bit length. can. Also, the remaining available resolutions can be indicated by indicator values expressed as '10' and '11', respectively.

On the other hand, if the resolution indicator is represented by bits of variable length, the signaling overhead is reduced when the available resolution indicated using the shortest length of bits is used to resolve the motion vector difference value of the current block. can be reduced. Accordingly, an available resolution with a high probability of being selected as the resolution of the motion vector difference value of the current block among the plurality of available resolutions can be set to be signaled using the shortest length bit. That is, the available resolutions indicated by the resolution indicators of the same indicator value may vary depending on the situation. This will be described in detail with reference to FIGS. 11 and 12. FIG. Also, the resolution set of the current block can be configured with an available resolution that is advantageous depending on the situation. That is, the configuration of the plurality of available resolutions included in the resolution set of the current block may change according to circumstances. This will be described in detail with reference to FIG.

11 and 12 are diagrams illustrating how the resolution of the motion vector difference value of the current block is signaled according to another embodiment of the present invention. Although FIG. 11 illustrates the resolution set as including three available resolutions, this disclosure is not so limited. In the embodiment of FIG. 11, the resolution set may include a first available resolution (Resolution 1), a second available resolution (Resolution 2), and a third available resolution (Resolution 3). Also, the value of the resolution indicator indicating each of the plurality of available resolutions included in the resolution set may be any one of 0, 10, and 11. FIG.

At this time, the usable resolution indicated by the resolution indicators of the same indicator value may vary depending on the situation. For example, the indicator value that uses the shortest bits among the resolution indicator values indicates the first available resolution in the first situation (case 1), and the second available resolution in the second situation (case 2). can give instructions. The first available resolution and the second available resolution may be different from each other.

Referring to 'case 1' of FIG. 11, the first available resolution (Resolution 1), the second available resolution (Resolution 2), and the third available resolution (Resolution 3) have resolution indicator values of 0, 10, and 10, respectively. 11 may be the available resolution indicated. 11, the first available resolution (Resolution 1), the second available resolution (Resolution 2), and the third available resolution (Resolution 3) are each set to the value of the resolution indicator. may be the available resolutions indicated when is 10, 0, 11;

FIG. 12 more specifically illustrates the embodiment of FIG. Depending on the situation, the resolution of the motion vector difference value of the current block can be signaled by the method of FIG. 12(a) or the method of FIG. 12(b). In the embodiment of FIG. 12, the resolution set can include available resolutions of 1/4, 1, and 4 samples. Also, each available resolution can be indicated by a different indicator value depending on the situation. For example, referring to FIG. 12(a), available resolutions of 1/4, 1, and 4 sample units can be signaled by 10, 0, 11, respectively. That is, the smallest available resolution among the plurality of available resolutions can be indicated by an indicator value that uses the shortest length of bits. This signaling method can be used in situations where it is advantageous to signal the motion vector difference value of the current block based on the smallest available resolution among a plurality of available resolutions.

Also referring to FIG. 12(b), available resolutions of 1/4, 1 and 4 sample units can be signaled by 10, 11 and 0, respectively. That is, according to an embodiment of the present invention, it may be advantageous to signal the motion vector difference value of the current block based on an available resolution that is not the smallest available resolution among a plurality of available resolutions depending on circumstances. In this case, the available resolution that is not the smallest available resolution among the plurality of available resolutions can be indicated by an indicator value that uses the shortest length bits. For example, the largest available resolution of a plurality of available resolutions can be indicated by an indicator value that uses the shortest length bits.

Also, the configuration of the plurality of available resolutions included in the resolution set can change depending on the situation. FIG. 13 is a diagram illustrating an embodiment in which the configuration of available resolutions included in a resolution set changes. According to one embodiment of the present invention, in a first situation (case 1) the resolution of motion vector difference values is obtained from the first situation resolution set, and in a second situation (case 2) the resolution of motion vector difference values is obtained from the second situation Can be obtained from resolution sets.

According to one embodiment, the first context resolution set may include a first available resolution (Resolution 1), a second available resolution (Resolution 2), and a third available resolution (Resolution 3). Also, the second context resolution set can include a fourth available resolution (Resolution A), a fifth available resolution (Resolution B), and a sixth available resolution (Resolution C). At this time, some of the available resolutions included in the first context resolution set and some of the available resolutions included in the second context resolution set may be the same. That is, the specific available resolution may be included in both the first context resolution set and the second context resolution set.

According to a specific embodiment, the first context resolution set can consist of available resolutions of 1/4, 1 and 4 samples. Also, the second context resolution set can consist of available resolutions in units of 1/4, 1/2, and 1 sample. That is, the second context resolution set can include other available resolutions instead of the largest available resolution among the plurality of available resolutions included in the first context resolution set. At this time, other available resolutions may be available resolutions smaller than the largest available resolution. Also, the largest available resolution among the available resolutions included in the second context resolution set may be smaller than the largest available resolution among the available resolutions included in the first context resolution set. According to one embodiment, the first context resolution set may be advantageous in situations requiring reduced signaling overhead compared to situations where the accuracy of motion vector difference values should be increased. Conversely, the second context resolution set may be advantageous in situations where the accuracy of motion vector difference values should be increased compared to situations where reduced signaling overhead is required.

Meanwhile, the accuracy of the motion vector difference value required may vary according to the prediction mode of the current block. For example, motion compensation based on affine models can be motion estimation for irregular motion other than translational motion. Accordingly, when the prediction mode of the current block is the motion compensation mode based on the affine model, it may be necessary to signal the motion vector difference value more precisely than in the existing general inter prediction mode. The MVP candidate list of the current block can be constructed in different ways according to the prediction mode of the current block. Accordingly, the configuration of the plurality of available resolutions included in the resolution set of the current block can vary depending on how the MVP candidate list for motion compensation of the current block is constructed.

Also, as the resolution of the motion vector difference value of the current block increases, the size of the signaled motion vector difference value can be reduced. Accordingly, when the absolute value of the motion vector difference value of the current block is relatively large, it is preferable to signal the motion vector difference value in a larger resolution unit than when the motion vector difference value of the current block is small. be. At this time, the motion vector difference value is a value indicating the difference between the motion vector and the motion vector predictor. Therefore, the more similar the motion vector predictor of the current block to the motion vector of the current block, the smaller the absolute value of the motion vector difference of the current block. On the other hand, the probability of high similarity between the motion vector of the current block and the motion vector predictor can vary depending on how the MVP candidate list for motion compensation of the current block is constructed. Accordingly, the configuration of the plurality of available resolutions included in the resolution set of the current block can vary depending on how the MVP candidate list for motion compensation of the current block is constructed.

According to one embodiment, the situations referred to in the above-described embodiments through FIGS. 10 to 13 may refer to methods in which the MVP candidate list of the current block is constructed. As described above with reference to FIG. 8, the MVP candidate list for motion compensation of the current block can be constructed based on any one of a plurality of methods. The MVP candidate list of the current block can be constructed in different ways depending on the prediction mode of the current block. That is, the situations referred to in the above embodiments can be defined as depending on which of a plurality of methods is used to construct the MVP candidate list for motion compensation of the current block.

According to an embodiment of the present invention, the configuration of multiple available resolutions included in the resolution set of the current block may vary depending on how the MVP candidate list of the current block is constructed. For example, the encoder and decoder can construct an MVP candidate list for motion compensation of the current block using any one of the first method and the second method described above. In this case, the configuration of the plurality of available resolutions included in the resolution set of the current block may vary depending on which of the first method and the second method is used to construct the MVP candidate list of the current block. can be done.

According to one embodiment, when the MVP candidate list for motion compensation of the current block is constructed using the first method, the resolution of motion vector difference values of the current block can be obtained from the first resolution set. When the MVP candidate list for motion compensation of the current block is constructed using the second method, the resolution of motion vector difference values of the current block can be obtained from the second resolution set. At this time, some of the plurality of available resolutions forming the first resolution set and some of the plurality of available resolutions forming the second resolution set may be different from each other. For example, the second resolution set may include at least one other available resolution than the plurality of available resolutions included in the first resolution set.

As described above, the first method may be a method of constructing the MVP candidate list based on the affine model, and the second method may be a method of constructing the MVP candidate list not based on the affine model. Affine model-based motion compensation can be motion estimation for irregular motion other than translational motion. Accordingly, it may be necessary to signal motion vector difference values more precisely than in existing general inter-prediction methods.

Accordingly, the first resolution set can include other available resolutions instead of the largest available resolution among the plurality of available resolutions included in the second resolution set. At this time, other available resolutions may be available resolutions smaller than the largest available resolution. Also, the largest available resolution among the available resolutions included in the first resolution set may be smaller than the largest available resolution among the available resolutions included in the second resolution set. For example, the largest available resolution among the available resolutions included in the first resolution set may be a resolution of 1 sample unit. At this time, the largest available resolution among the available resolutions included in the second resolution set may be a 4-sample unit resolution.

According to a specific embodiment, the first resolution set can consist of available resolutions of 1/4, 1/2 and 1 sample units. At this time, the second resolution set can be composed of available resolutions of 1/4, 1, and 4 samples. According to another specific embodiment, the first resolution set can consist of available resolutions of 1/16, 1/4 and 1 sample units. At this time, the second resolution set can be composed of available resolutions of 1/4, 1, and 4 samples.

As described above, the decoder can determine the resolution of motion vector difference values of the current block using the signaled resolution indicator. For example, the decoder can obtain a resolution indicator indicating the resolution of the motion vector difference value of the current block among a plurality of available resolutions included in one of the first resolution set and the second resolution set. Also, the decoder can obtain a modified motion vector difference value of the current block based on the resolution indicator. A decoder can reconstruct the current block based on the modified motion vector difference values.

According to one embodiment of the present invention, the available resolution indicated by a particular one of the resolution indicator values depends on which of the first method and the second method the MVP candidate list is constructed using. can change. According to one embodiment, a decoder can obtain a resolution indicator having a first value. If the MVP candidate list for motion compensation of the current block is constructed using the first method, the available resolution indicated by the first value may be the seventh available resolution. If the MVP candidate list for motion compensation of the current block is constructed using the second method, the available resolution indicated by the first value may be the eighth available resolution. At this time, the seventh available resolution and the eighth available resolution may be different resolutions.

According to one embodiment, the seventh available resolution may be the available resolution included in both the first resolution set and the second resolution set. If the MVP candidate list for motion compensation of the current block is constructed using the second method, the seventh available resolution can be indicated by a second value different from the first value. For example, the first value may be one of '10, 11' or one of '00, 01'. At this time, the second value may be one of '10 and 11' different from the first value, or one of '00 and 01' different from the first value.

On the other hand, as described above, it is advantageous to use bits with the shortest length to signal the available resolution with the highest probability of being selected as the resolution of the motion vector difference value of the current block among the plurality of available resolutions. Possible. According to one embodiment, the smallest available resolution of the first resolution set may be smaller than the smallest available resolution of the second resolution set.

That is, the available resolutions included in the second resolution set may be relatively large resolutions. Accordingly, if the MVP candidate list for motion compensation of the current block is constructed using the second method, it may be necessary to increase the precision of the motion vector difference value. As previously mentioned, if there is a need to increase the accuracy of motion vector difference values, it may be advantageous to signal the smallest available resolution of multiple available resolutions using the shortest length bits. According to one embodiment, the third value may be a value expressed using the shortest bits among the values of the resolution indicator.

According to one embodiment, when the MVP candidate list for motion compensation of the current block is constructed using the second method, the third value is the smallest available resolution among the plurality of available resolutions included in the second resolution set. can be instructed. According to one embodiment, the smallest available resolution among the available resolutions included in the second resolution set may be 1/4.

Also, the available resolutions included in the first resolution set may be relatively small resolutions. Accordingly, if the MVP candidate list for motion compensation of the current block is constructed using the first method, it may be necessary to reduce the signaling overhead for motion vector difference values. Therefore, when the MVP candidate list for motion compensation of the current block is constructed using the first method, the third value is a value other than the smallest available resolution among the plurality of available resolutions included in the first resolution set. Other available resolutions can be indicated. According to one embodiment, the third value may indicate the largest available resolution, the second smallest available resolution, or the second largest available resolution among the available resolutions included in the first resolution set. For example, the available resolutions other than the smallest available resolution among the plurality of available resolutions included in the first resolution set may be any one of 1/4, 1/2, 1, and 4. FIG.

Meanwhile, according to another embodiment of the present invention, the configuration of available resolutions included in the resolution set of the current block may vary according to the MVP index indicating the motion vector predictor of the current block. The MVP index may indicate what order of the candidate in the MVP candidate list of the current block is the motion vector predictor referred to derive the motion vector of the current block.

FIG. 14 is a diagram illustrating an embodiment of how the resolution of the motion vector difference value of the current block is signaled by the motion vector predictor of the current block. For example, the decoder can obtain an MVP index indicating a motion vector predictor referenced for motion compensation of the current block among the MVP candidate list for motion compensation of the current block. At this time, the available resolution indicated by the specific value of the resolution indicator of the current block may vary depending on whether the MVP index is larger or smaller than the preset index.

A motion vector predictor candidate corresponding to a smaller MVP index in the MVP candidate list of the current block may have a higher probability of being similar to the motion vector of the current block. Also, the more similar the motion vector predictor of the current block to the motion vector of the current block, the smaller the absolute value of the motion vector difference of the current block. Accordingly, when the MVP index corresponding to the motion vector predictor of the current block is smaller than the preset MVP index, the smallest available resolution among the plurality of available resolutions included in the resolution set of the current block is the resolution indicator. It can be indicated by a value expressed using the bits with the shortest length of the value.

Conversely, if the MVP index corresponding to the motion vector predictor of the current block is greater than the preset MVP index, the resolution is a specific available resolution that is not the smallest available resolution among a plurality of available resolutions included in the resolution set of the current block. It can be indicated by a value represented using the shortest length bits of the indicator value. At this time, the specific available resolution may be the largest resolution among the plurality of available resolutions. Alternatively, the specific available resolution may be the second largest or second smallest resolution among the plurality of available resolutions.

Referring to FIG. 14, the current block resolution set may include available resolutions of 1/4, 1, and 4 samples. In addition, when the MVP index for motion compensation of the current block is smaller than the preset MVP index (Candidate 1, Candidate 2), the smallest available resolution among 1/4, 1, and 4 is the unit of 1/4. resolution can be signaled using the shortest bits. On the other hand, if the MVP index for motion compensation of the current block is greater than the preset MVP index (Candidate N), the least number of bits with a resolution of 1 or 4 units among 1/4, 1, and 4 is selected. can be used for signaling.

FIG. 15 is a diagram illustrating another embodiment of a method in which the motion vector predictor of the current block signals the resolution of the motion vector difference value. As described above with reference to FIG. 14, the method of signaling the resolution of the motion vector difference value of the current block may vary according to the MVP index for motion compensation of the current block.

As described above, the MVP candidate list for motion compensation of the current block can include candidates obtained by various methods. According to one embodiment, the MVP candidate list for motion compensation of the current block may include at least one of spatial candidates, temporal candidates, or zero motion vectors. At this time, the configuration of a plurality of available resolutions included in the resolution set of the current block can be changed depending on how the motion vector predictor of the current block is obtained as a candidate. This is because the similarity with the current block estimated by the motion vector predictor candidate may differ. For example, it can be estimated that spatial candidates have a high probability of being similar to the current block. In addition, it can be estimated that motion vector predictor candidates such as temporal candidates and zero motion vectors are less likely to be similar to the current block than spatial candidates.

For example, if the motion vector predictor of the current block is one of the spatial candidates, the smallest available resolution among the multiple available resolutions included in the resolution set of the current block is the value of the resolution indicator. It can be indicated by a value represented using the shortest length bit of them.

Also, if the motion vector predictor of the current block is not one of the spatial candidates, the specific available resolution that is not the smallest available resolution among the plurality of available resolutions included in the resolution set of the current block is the resolution indicator. It can be indicated by a value represented using the shortest length bits of the value. At this time, the specific available resolution may be the largest resolution among the plurality of available resolutions. Alternatively, the specified available resolution may be the second largest or second smallest resolution of the plurality of available resolutions.

On the other hand, although FIGS. 13-15 illustrate that each of the different resolution sets includes the same number of available resolutions, the present disclosure is not limited thereto. According to an embodiment of the present invention, each resolution set may include a different number of available resolutions. For example, the first resolution set may contain three available resolutions. On the other hand, the third resolution set used in the third situation can contain two available resolutions. The encoder and decoder may configure a resolution set including a different number of available resolutions depending on the situation.

As described above, the motion information set can include a reference picture index that indicates the reference picture of the current block. The decoder can obtain the POC of the reference picture of the current block from the signaled reference picture index. According to an embodiment of the present invention, the configuration of multiple available resolutions included in the resolution set of the current block may vary according to the POC of the current picture and the POC of the reference picture. According to one embodiment, based on the POC of the current picture and the POC of the reference picture, some of the available resolutions that make up the resolution set of the current block can be excluded.

For example, if the POC of the reference picture is the same as the POC of the current picture, the resolution of the motion vector difference value of the current block can be obtained from the fourth resolution set. Also, if the POC of the reference picture is not the same as the POC of the current picture, the resolution of the motion vector difference value of the current block can be obtained from the fifth resolution set. At this time, the fourth resolution set may be composed of the remaining available resolutions excluding the smallest available resolution among the available resolutions included in the fifth resolution set. That is, the number of available resolutions forming the resolution set of the current block may vary according to the POC of the current picture and the POC of the reference picture.

According to another embodiment of the present invention, the resolution of the motion vector difference value of the current block can be signaled in different ways according to a quantization parameter (QP). For example, one of the signaling methods described above with reference to FIGS. 11 to 15 can be determined according to the QP of the current block. Also, the resolution of the motion vector difference value of the current block can be signaled using the determined signaling method. According to one embodiment, the smaller the QP of the current block, the smaller the resolution of the motion vector difference value of the current block.

According to still another embodiment of the present invention, the resolution of the motion vector difference value of the current block can be signaled in different ways depending on the framerate. For example, one of the signaling methods described above with reference to FIGS. 11 to 15 can be determined according to the frame rate of the video signal including the current block. Also, the resolution of the motion vector difference value of the current block can be signaled using the determined signaling method. According to one embodiment, the higher the frame rate of the video signal, the higher the probability that the time interval between frames is short and the motion vector is small. Accordingly, the higher the frame rate of the current block, the smaller the resolution of the motion vector difference value of the current block.

According to yet another embodiment of the present invention, the resolution of the motion vector difference value of the current block can be signaled in different ways according to the size ratio or size difference between the current block and its neighboring blocks. For example, one of the signaling methods described above with reference to FIGS. 11 to 15 can be determined according to the size ratio or size difference between the current block and neighboring blocks of the current block. The above-described embodiments can be applied in the same or corresponding manner even when motion vectors other than motion vector differential values are directly signaled.

However, according to additional embodiments of the present invention, the encoder and decoder can modify the motion vector predictor before adding the motion vector difference values. Through this, it is possible to reduce the size of the motion vector differential value used to obtain the motion vector of the current block. At this time, the motion vector predictor can be modified only when some of the motion vector predictor candidates included in the MVP candidate list of the current block are selected as the motion vector predictor of the current block. For example, if a motion vector predictor candidate corresponding to an MVP index smaller than a preset MVP index among motion vector predictor candidates included in the MVP candidate list of the current block is used as the motion vector predictor of the current block, the corresponding Corrections to motion vector predictors can be performed.

According to one embodiment, the method of signaling the multiple available resolutions included in the resolution set may vary depending on whether a modification to the motion vector predictor is performed. For example, if the motion vector predictor of the current block is modified, the smallest available resolution among the plurality of available resolutions included in the resolution set can be signaled using the least number of bits. Also, if the motion vector predictor of the current block has not been modified, any one of the multiple available resolutions included in the resolution set that is not the smallest available resolution uses the least number of bits. can be signaled.

According to another embodiment, the motion vector predictor can also be modified after summing the motion vector difference values. In this case, the resolution of the motion vector difference values of the current block can be signaled in a different way than if the motion vector predictor were modified before summing the motion vector difference values. If the motion vector predictor is corrected after the motion vector difference value is added, the prediction error can be reduced through the motion vector correction process even if the motion vector difference value has low accuracy. As a result, if the motion vector predictor of the current block is modified before adding the motion vector difference value, any one of the other resolutions that is not the smallest available resolution among the multiple available resolutions included in the resolution set is It can be signaled using the fewest number of bits. Conversely, if the motion vector predictor of the current block is not modified before adding the motion vector difference value, the smallest available resolution among the multiple available resolutions included in the resolution set can be signaled using the least number of bits. .

According to an additional embodiment, the motion vector correction process to be performed later may vary depending on what candidate the motion vector or motion vector predictor of the current block is. The process of correcting motion vectors can include processes for searching for more accurate motion vectors. For example, the decoder can search for a block matching the current block from a reference point by a predefined method. The reference point may be a position corresponding to the determined motion vector or motion vector predictor. Motion vector search can be performed in a variety of ways. For example, template matching or bilateral matching can be used for searching. At this time, the degree of movement from the reference point can be changed according to a predefined method. Different motion vector correction processes can be different degrees of movement from the reference point.

For example, an accurate motion vector predictor candidate may start with a detailed correction process, and an inaccurate motion vector predictor candidate may start with a less detailed correction process. In this disclosure, the correct motion vector predictor candidate and the incorrect motion vector predictor candidate can be determined by their position within the MVP candidate list. Also, an accurate motion vector predictor candidate and an inaccurate motion vector predictor candidate can be determined according to how each motion vector predictor candidate is generated. The method by which the motion vector predictor candidate is generated can indicate which position the candidate corresponds to among the spatial candidates of the current block. Also, detailed and less detailed modifications can be whether the block is explored in small increments or large increments from the reference point. In addition, when searching while moving a lot, it is possible to add a process of searching while moving little by little from the best matching block found while moving a lot.

The template matching method described above may be a method of obtaining a comparison target block having the smallest difference from the current block template based on the difference in value between the current block template and the comparison target block template to be compared. A template for a particular block can be obtained based on the surrounding samples of the particular block. FIG. 16 is a diagram illustrating how the resolution of motion vector difference values is obtained based on the template matching method according to an embodiment of the present invention. According to one embodiment of the present invention, the resolution of motion vector difference values of the current block may not be signaled. According to one embodiment, the decoder can obtain the resolution of the current block based on the same cost calculation as the template matching method. As described above, reference blocks can be obtained based on motion vector predictors and motion vector difference values. At this time, the motion vector difference value may be a motion vector difference value corrected according to the resolution of the motion vector difference value of the current block. Accordingly, the position of the reference block within the reference picture can be changed according to the resolution of the motion vector difference value of the current block.

According to an embodiment, the resolution set for motion compensation of the current block can be configured with preset available resolutions. Referring to FIG. 16, the resolution set of the current block may include a first available resolution (Resolution 1), a second available resolution (Resolution 2), and a third available resolution (Resolution 3). A plurality of reference block candidates corresponding to each of the available resolutions included in the resolution set of the current block can be obtained based on the reference point 1601 indicated by the motion vector predictor of the current block. As many reference block candidates as the number of available resolutions included in the resolution set can be obtained. According to a specific embodiment, the plurality of reference block candidates are a first reference block candidate 1602 corresponding to a first available resolution, a second reference block candidate 1603 corresponding to a second available resolution, and a third reference block candidate corresponding to a third available resolution. 3 reference block candidates 1604 may be included. A decoder can use any one of a plurality of reference block candidates as a reference block for the current block. For example, the decoder can select the reference block candidate with the lowest cost as the reference block of the current block based on the template matching result between each of the plurality of reference block candidates and the current block. Also, the decoder can reconstruct the current block based on the corresponding reference block.

According to another embodiment, the resolution of the motion vector difference value of the current block may be signaled based on the template matching result between each of the reference block candidates and the current block. For example, the template matching operation described above can be performed in the same manner in the encoder as in the decoder. Also, the encoder can use the fewest number of bits to signal the available resolution corresponding to the lowest cost reference block candidate based on the template matching results.

In this case, the encoder and decoder may perform template matching on only some of the plurality of reference block candidates. For example, among the first reference block candidate 1602, second reference block candidate 1603, and third reference block candidate 1604 in FIG. Template matching can be performed. Also, based on the template matching result between each of the first reference block candidate 1602 and the second reference block candidate 1603 and the current block, among the available resolutions included in the resolution set of the current block, bits with the shortest length are selected. The available resolution signaled using can be determined. Either one of the first available resolution and the second available resolution can be signaled using the shortest length bits. In addition, the remaining one of the first available resolution and the second available resolution and the third available resolution corresponding to the reference block candidate for which template matching is not performed may be signaled using an additional bit. Through this, the encoder and decoder can reduce the amount of computation required for the template matching method.

Whether or not to apply the embodiment described through FIG. 16 according to one embodiment can also be determined according to the size of the signaled motion vector difference value. For example, the template matching method can be used only when the signaled motion vector difference value is greater than a preset value. This is because if the signaled motion vector difference value is smaller than a preset value, the template matching cost difference between the reference block candidates according to the resolution may not be obvious.

FIG. 17 is a diagram illustrating how the resolution of motion vector difference values is obtained based on the two-sided matching method according to an embodiment of the present invention. According to an embodiment of the present invention, a two-sided matching method can be used instead of the template matching method described with reference to FIG. A two-sided matching method refers to a method of reconstructing a current block based on reference blocks in each of two or more reference pictures along a motion trajectory. A two-sided matching method may be one that obtains the smallest set of differences based on differences between reference blocks in each of two or more reference pictures.

As described above, if the current block is a bi-predictive block, the current block can be reconstructed based on two or more reference blocks of two or more different reference pictures. Referring to FIG. 17, reference block candidates corresponding to a specific available resolution can be configured for each of a first reference picture (Reference picture 1) and a second reference picture (Reference picture 2). The encoder and decoder select the first reference picture (Reference picture 1) based on the two-sided matching result between the reference block candidate in the first reference picture (Reference picture 1) and the reference block candidate in the second reference picture (Reference picture 2). 1) and the reference block in the second reference picture (Reference picture 2) can be obtained. The embodiment described above through FIG. 16 can be applied to the embodiment of FIG. 17 in the same or corresponding manner.

According to one embodiment, when the current block is inter-predicted based on two motion information sets corresponding to different reference lists, two motion vector difference values can be separately signaled. In this case, the resolutions applied to each of the two motion vector difference values may be the same or different. Various methods of signaling the resolution of the current block by reference picture list will be described below.

According to an embodiment of the present invention, the resolution of the motion vector difference value for each reference picture list can be determined based on the two-sided matching method described above. According to one embodiment, resolution sets corresponding to each of the two reference picture lists can be configured independently of each other. For example, the resolution set corresponding to the first reference picture list L0 may contain m available resolutions. Also, the resolution set corresponding to the second reference picture list L1 may include n available resolutions. In this case, the encoder and decoder each of the reference picture lists based on the two-sided matching results between the n reference block candidates corresponding to the first reference picture list and the m reference block candidates corresponding to the second reference picture list. motion vector difference value resolution can be obtained. In this case, the encoder and decoder must perform (nxm) two-sided matching.

Alternatively, a common resolution set can be used for a plurality of reference picture lists as the resolution of the motion vector difference value. For example, a commonly used resolution set for the first reference picture list L0 and the second reference picture list L1 may include n available resolutions. According to an embodiment, the same resolution may be applied to motion vector difference values corresponding to each of the first reference picture list L0 and the second reference picture list L1. In this case, the encoder and decoder can obtain resolutions of motion vector difference values of each of the first reference picture list L0 and the second reference picture list L1 based on the n-times two-sided matching results.

Meanwhile, a difference between a motion vector predictor and a motion vector corresponding to a particular reference picture list may be similar to a difference between a motion vector predictor and a motion vector corresponding to another reference picture list. In this case, the resolution of motion vector difference values corresponding to a particular reference picture list may be likely to be the same as the resolution of motion vector difference values corresponding to other reference picture lists. Therefore, according to an embodiment of the present invention, the resolution of motion vector difference values corresponding to a particular reference picture list of the current block is the same as the resolution of motion vector difference values corresponding to other reference picture lists of the current block. Possible.

FIG. 18 illustrates a method of signaling the resolution of a motion vector difference value for each reference picture list of a current block according to an embodiment of the present invention. Referring to FIG. 18 , a first list resolution set (L0 MVD resolution signaling) corresponding to a first reference picture list L0 and a second list resolution set (L1 MVD resolution signaling) corresponding to a second reference picture list L1 are respectively configured. can. A first list resolution indicator that indicates the resolution of the motion vector difference value corresponding to the first reference picture list L0 and a second list resolution indicator that indicates the resolution of the motion vector difference value corresponding to the second reference picture list L1. Each child can signal.

For example, the decoder can obtain the resolution of motion vector difference values corresponding to the first reference picture list based on the first list resolution set and the first list resolution indicator. Next, the decoder can obtain the resolution of the motion vector difference values corresponding to the second reference picture list based on the second list resolution set and the second list resolution indicator. At this time, one of the available resolutions included in the second list resolution set may be a resolution (L0 resolution) depending on the resolution of the motion vector difference value corresponding to the first reference picture list.

According to a specific embodiment, if the second list resolution indicator is a preset value (0), the decoder may select the second reference picture list based on the resolution of the motion vector difference values corresponding to the first reference picture list. can determine the resolution of the motion vector difference values corresponding to . In this case, the decoder can use the same resolution as the resolution of the motion vector differential values corresponding to the first reference picture list for the resolution of the motion vector differential values corresponding to the second reference picture list. At this time, the preset value may be a value expressed using the least number of bits among the values of the second list resolution indicator. Also, other values of the second list resolution indicator can be used as indicator values respectively indicating the remaining available resolutions (Remaining resolution 1, Remaining resolution 2).

Although it has been described in the embodiment of FIG. 18 that the resolution of the motion vector difference values corresponding to the first reference picture list L0 is determined prior to the resolution of the motion vector difference values corresponding to the second reference picture list L1. , the disclosure is not limited thereto. For example, the first list resolution set can contain resolutions that depend on the resolution of the motion vector difference values corresponding to the second reference picture list L1.

FIG. 19 is a diagram illustrating an embodiment of how the resolution of the motion vector difference value of the current block is signaled by the picture resolution. According to an embodiment of the present invention, the resolution of the motion vector difference value of the current block can be signaled in different ways depending on at least one of picture resolution and size. Information about picture resolution or size can be signaled from the encoder. The decoder can obtain information about the resolution or size of the signaled picture. According to an embodiment, a method of signaling the resolution of a motion vector difference value of a specific block in each of a low resolution picture and a high resolution picture may be different. Referring to FIG. 19, when the relative distance from the reference point indicated by the motion vector predictor to the position of the reference block is the same in each of the low resolution picture and the high resolution picture. However, the value for indicating the corresponding distance in the high-resolution picture may be larger than the value for indicating the corresponding distance in the low-resolution picture.

Accordingly, the resolution of the motion vector difference value of the current block is likely to be set to a value larger than the resolution of the motion vector difference value of another picture whose resolution is lower than that of the current picture including the current block. be. Therefore, if the current picture is a high resolution picture, the smallest available resolution of the multiple available resolutions may not be indicated by the value represented using the shortest length bits. For example, if the current picture is a high resolution picture, the available resolution that is not the smallest available resolution among the plurality of available resolutions can be indicated by a value expressed using the shortest length bits. The above-described embodiments can also be applied when the resolution and size of pictures included in a video signal change, such as in scalable video coding.

FIG. 20 illustrates how the resolution of the motion vector difference value is signaled based on the size of the reference picture of the current block according to one embodiment of the present invention. According to one embodiment, the resolution of motion vector difference values can be signaled based on the size of the reference picture of the current block. FIG. 20 shows reference block candidates corresponding to each of the available resolutions included in the resolution set of the current block based on the reference point 2001 indicated by the motion vector predictor of the current block. In FIG. 20, the third reference block candidate corresponding to the third available resolution (Resolution 3) is located outside the reference picture. As such, when at least part of reference block candidates configured based on a specific available resolution is outside the reference picture boundary, the available resolution can be excluded from signaling. That is, if the point indicated by the motion vector difference value modified based on the specific available resolution with reference to the reference point 2001 is outside the boundary of the current picture, the available resolution may not be signaled.

For example, a resolution set may contain N available resolutions. In this case, a resolution indicator can be signaled to indicate one of (NM) excluding M available resolutions among N available resolutions. At this time, reference block candidates corresponding to each of the M available resolutions may be reference block candidates outside the boundary of the reference picture. Through this, the encoder and decoder can reduce the signaling overhead for the resolution of the motion vector difference value of the current block by excluding unnecessary available resolutions. According to a specific example, when signaling any one of the first available resolution (Resolution 1), the second available resolution (Resolution 2), and the third available resolution (Resolution 3), maximum 2 bits are used. be able to. Meanwhile, when signaling one of the first available resolution (Resolution 1) and the second available resolution (Resolution 2), a maximum of 1 bit may be used.

Referring to FIG. 20, a resolution set can include a first available resolution, a second available resolution, and a third available resolution. At this time, if the signaled motion vector difference value of the current block is modified based on the third available resolution, the motion vector of the current block points to a point outside the boundary of the reference block. Accordingly, the encoder can signal a resolution indicator indicating one of the first available resolution and the second available resolution excluding the third available resolution. According to a specific embodiment, the encoder can use one bit to signal a resolution indicator that indicates one of the first available resolution and the second available resolution.

Also, the decoder can determine available resolutions to be excluded from signaling based on the size of the reference picture of the current block and the signaled motion vector difference value. In the embodiment of FIG. 20, the decoder can recognize that the third available resolution is not signaled. A decoder can obtain the resolution of the motion vector difference value of the current block based on the resolution indicator signaled through one bit.

According to an additional embodiment of the present invention, it is possible to decide whether to use the method described above with reference to FIG. 20 based on the position of the reference point indicated by the motion vector predictor. For example, if the position of the reference point is within a preset distance from the boundary of the reference picture, the method described with reference to FIG. 20 can be used.

Whether or not to apply the embodiment described through FIG. 20 according to one embodiment can be determined according to the size of the signaled motion vector difference value. For example, only if the signaled motion vector difference value is greater than a preset value can the encoder and decoder consider if there are any available resolutions to generate reference block candidates that deviate from the reference picture. If the signaled motion vector difference value is smaller than a preset value, the encoder and decoder may not use the method described with reference to FIG.

According to additional embodiments, new available resolutions can be used in place of available resolutions that have been excluded from signaling. For example, if the reference block candidate corresponding to the motion vector difference value modified based on the specific available resolution is located outside the boundary of the reference picture, the indicator value indicating the specific available resolution is different from the specific available resolution. can be instructed. For example, different available resolutions may be less than a particular available resolution. According to a specific embodiment, the resolution set of the current block can consist of available resolutions of 1/4, 1 and 4 samples. At least some of the reference block candidates corresponding to the available resolution of 4 samples may be located outside the boundary of the reference picture of the current block. In this case, the resolution indicator value indicating the available resolution in units of 4 samples may indicate the available resolution in units of 2 samples.

FIG. 21 is a flowchart illustrating a method for obtaining the resolution of the motion vector difference value of the current block according to one embodiment of the present invention. Although FIG. 21 illustrates the resolution set for the current block as including available resolutions of 1/4, 1, and 4 samples, this disclosure is not so limited. For example, at least some of the available resolutions of 1/4, 1, and 4 sample units can be replaced with some of the available resolutions of 1/2, 2, 1/8, and 1/16 sample units.

Referring to FIG. 21, the decoder can obtain the signaled motion vector difference value of the current block by parsing the bitstream (S2101). At this time, the signaled motion vector difference value may be a value expressed in units of resolution of the motion vector difference value.

Also, the decoder can obtain a resolution indicator that indicates the resolution of the motion vector difference value of the current block. According to one embodiment, the resolution indicator can indicate any one of multiple values represented by bits of variable length. The decoder can parse the first bit of the resolution indicator (S2103). Next, the decoder can determine whether the first bit of the resolution indicator is '0' (S2105). If the first bit is '0', the available resolution of 1/4 units can be used for resolution of the motion vector difference value of the current block. In this case, the resolution indicator may not include the second bit described below.

If the first bit is not '0', the decoder can determine whether the reference block candidate corresponding to the 4-sample unit available resolution is located within the boundary of the reference picture (S2107). Here, the reference block candidate corresponding to the available resolution of 4 samples may be a reference block candidate indicated by the motion vector candidate obtained based on the available resolution of 4 samples. Specifically, the motion vector candidate is a value obtained by adding a motion vector difference value corrected based on the available resolution in units of 4 samples from the signaled motion vector difference value in step S2101 to the motion vector predictor of the current block. Possible. According to one embodiment, if the reference block candidate corresponding to the available resolution of 4 samples is not located within the boundary of the reference picture, the available resolution of 1 sample can be used as the resolution of the motion vector difference value of the current block. . In this case, the resolution indicator may not include the second bit described below.

If the reference block candidate corresponding to the available resolution of 4 sample units is located within the boundary of the reference picture, the decoder can parse the second bit of the resolution indicator (S2109). Next, the decoder can determine the resolution of the motion vector difference value of the current block based on the second bit of the resolution indicator (S2111). For example, the indicator value indicating the available resolution in units of 1 sample may be '10', and the indicator value indicating the available resolution in units of 4 samples may be '11'. If the second bit of the resolution indicator is '0', the decoder can use the available resolution of 1 sample unit as the resolution of the motion vector difference value of the current block. Conversely, if the second bit of the resolution indicator is '1', the decoder can use the available resolution of 4 samples as the resolution of the motion vector differential value of the current block.

Although the step of obtaining the signaled motion vector difference value (S2101) is shown in FIG. not restricted. For example, step S2101 may follow step S2103. According to one embodiment, when the resolution indicator is represented by a plurality of bits, each bit can be implemented as a separate index or flag on the syntax for video signal processing.

According to additional embodiments, the step of parsing the second bit of the resolution indicator (S2109) can be performed at least after the step of obtaining the motion vector predictor of the current block. It is possible to determine whether to parse the second bit of the resolution indicator according to the position indicated by the motion vector predictor of the current block. This is because, as described above, when a reference block candidate corresponding to a specific available resolution is located outside the boundary of a reference picture, signaling for the corresponding available resolution can be omitted. That is, the second bit of the resolution indicator can be parsed after the motion vector predictor information of the current block in the syntax for encoding or decoding the video signal.

FIG. 22 is a flowchart illustrating a method for obtaining the resolution of the motion vector difference value of the current block according to one embodiment of the present invention. Referring to FIG. 22, the decoder can parse the bitstream to obtain the signaled motion vector difference value of the current block ( S 2 2 01). At this time, the signaled motion vector difference value may be a value expressed in units of resolution of the motion vector difference value. Also, the decoder can obtain a resolution indicator that indicates the resolution of the motion vector difference value of the current block. The decoder can parse the nth bit of the resolution indicator (S2205). Here, n can be an integer from 1 to N. n may be a value that is sequentially increased from 1 to N through a loop operation. If the nth bit of the resolution indicator is '0', the decoder can use the nth available resolution for the resolution of the motion vector difference value of the current block. At this time, the n-th available resolution may be the available resolution indicated by the n-th indicator value among the available resolutions included in the resolution set of the current block.

FIG. 22 is a flowchart illustrating a method for obtaining the resolution of the motion vector difference value of the current block according to one embodiment of the present invention. Referring to FIG. 22, the decoder can parse the bitstream to obtain the signaled motion vector difference value of the current block (2001). At this time, the signaled motion vector difference value may be a value expressed in units of resolution of the motion vector difference value. Also, the decoder can obtain a resolution indicator that indicates the resolution of the motion vector difference value of the current block. The decoder can parse the nth bit of the resolution indicator (S2205). Here, n can be an integer from 1 to N. n may be a value that is sequentially increased from 1 to N through a loop operation. If the nth bit of the resolution indicator is '0', the decoder can use the nth available resolution as the resolution of the motion vector difference value of the current block. At this time, the n-th available resolution may be the available resolution indicated by the n-th indicator value among the available resolutions included in the resolution set of the current block.

If the nth bit of the resolution indicator is not '0', the decoder can exclude the nth available resolution from the previous resolution set (S2207). The decoder may obtain a first current resolution set by excluding the n-th available resolution from the previous resolution set (S2207). According to one embodiment, if n is 1, the decoder may exclude the nth available resolution from the initial resolution set to obtain a first current resolution set. Next, the decoder can determine whether the number of available resolutions included in the first current resolution set is one (S2209). If the number of available resolutions included in the first current resolution set is one, the decoder can use the available resolutions included in the first current resolution set as the resolution of the motion vector difference value of the current block.

Next, the decoder can determine whether the number of available resolutions included in the second current resolution set is one (S2213). If the number of available resolutions included in the second current resolution set is one, the decoder can use the available resolutions included in the second current resolution set as the resolution of the motion vector difference value of the current block. If the number of available resolutions included in the first current resolution set is not one , the decoder may also perform steps S2203 through S2213 . Also, the decoder can increment n by one. At this time, the decoder can use the second current resolution set as the previous resolution set in step S2207 of the next loop operation.

Next, the decoder can determine whether the number of available resolutions included in the second current resolution set is one (S2213). If the number of available resolutions included in the second current resolution set is one, the decoder can use the available resolutions included in the second current resolution set as the resolution of the motion vector difference value of the current block. If the number of available resolutions included in the first current resolution set is not 1, the decoder may also perform steps 2203 through 2013 . Also, the decoder can increment n by one. At this time, the decoder can use the second current resolution set as the previous resolution set in step S2207 of the next loop operation.

FIG. 23 is a flow chart illustrating how the motion vector of the current block is obtained according to one embodiment of the present invention. According to an embodiment of the present invention, the information indicating the motion vector difference value of the current block may include first information about the motion vector difference value and second information about the motion vector difference value. At this time, the first information and the second information may be different information about the motion vector difference value. Referring to FIG. 23, the decoder can parse the first information for motion vector difference values. Also, the decoder can parse the second information for the motion vector difference values. Next, the decoder can obtain the motion vector of the current block based on the first information about the motion vector difference value and the second information about the motion vector difference value. According to an embodiment of the present invention, any one of the steps of parsing the first information or the second information for the motion vector difference value in FIG. 23 may be omitted. Hereinafter, conditions for parsing the first information or the second information about the motion vector difference value of the current block will be described.

FIG. 24 is a flow chart illustrating how the motion vector of the current block is obtained according to one embodiment of the present invention. According to an embodiment of the present invention, the information indicating the motion vector difference value of the current block may include first information about the motion vector difference value and second information about the motion vector difference value. In a specific situation, the motion vector of the current block can be determined through one of the first information about the motion vector difference value and the second information about the motion vector difference value. Meanwhile, in other situations, it may be difficult to determine one motion vector using only one of the first information about the motion vector difference value and the second information about the motion vector difference value.

Referring to FIG. 24, the decoder can parse the first information for motion vector difference values. Next, the decoder can obtain a plurality of reference block candidates using the first information on the motion vector difference values. The decoder can determine whether all of the multiple reference block candidates lie within the boundaries of the reference picture. If some of the plurality of reference block candidates are located outside the boundary of the reference picture, the decoder can obtain the motion vector of the current block based on the first information on the motion vector difference value. The decoder can parse the second information for motion vector difference values if all of the plurality of reference block candidates are located within the boundary of the reference picture. Next, the decoder can obtain the motion vector of the current block based on the first information about the motion vector difference value and the second information about the motion vector difference value.

According to an embodiment, the first information about the motion vector difference value of the current block may be information excluding sign information of the motion vector difference value. For example, the first information about the motion vector difference value of the current block may indicate the absolute value of the motion vector difference value of the current block. Also, the second information about the motion vector difference value of the current block may indicate the sign of the motion vector difference value of the current block. According to the above-described embodiments, a motion vector difference value of a specific block can be obtained without parsing the second information. This allows the encoder and decoder to reduce signaling overhead for motion vector differential values. Hereinafter, a method of implicitly signaling the sign information of the motion vector difference value of the current block will be described in detail.

Referring to FIG. 25, the x-axis component of the motion vector difference value of the current block with a negative sign (-) and the y-axis component of the motion vector difference value of the current block with a positive sign (+) are applied. Based on this, a fourth reference block candidate (Candidate reference block 1) can be obtained. Also, based on the x-axis component of the motion vector difference value of the current block to which the positive sign ( + ) is applied and the y-axis component of the motion vector difference value of the current block to which the positive sign (+) is applied, 5 reference block candidates (Candidate reference block 2) can be obtained.

Referring to FIG. 25, the x-axis component of the motion vector difference value of the current block with a negative sign (-) and the y-axis component of the motion vector difference value of the current block with a positive sign (+) are applied. Based on this, a fourth reference block candidate (Candidate reference block 1) can be obtained. Also, based on the x-axis component of the motion vector difference value of the current block to which the positive sign (+) is applied and the y-axis component of the motion vector difference value of the current block to which the positive sign (-) is applied, 5 reference block candidates (Candidate reference block 2) can be obtained.

According to one embodiment, if any one of the fourth candidate reference block (Candidate reference block 1) and the fifth candidate reference block (Candidate reference block 2) is located outside the boundary of the reference picture, the current Sign information of the x-axis component of the block motion vector difference value can be hidden. In this case, the encoder may not encode the hidden code information. In addition, the decoder can determine the sign of the motion vector difference value corresponding to the hidden code information without parsing the corresponding code information.

In FIG. 25, the fourth reference block candidate (Candidate reference block 1) can be located outside the boundary of the reference picture. In this case, sign information of the x-axis component of the motion vector difference value of the current block can be hidden. The decoder can obtain the sign information of the x-axis component of the motion vector differential value without parsing the sign information of the x-axis component of the motion vector differential value. A decoder can use a positive sign for the sign of the x-axis component of the motion vector difference value.

FIG. 26 illustrates how the sign bit of the motion vector difference value of the current block is implicitly signaled according to an embodiment of the present invention. According to one embodiment, the motion vector predictor for the current block may not be determined until the sign of the motion vector difference value for the current block is determined. For example, the sign of the motion vector difference value of the current block can be determined before the MVP index, which indicates the MVP of the current block on the syntax for video signal processing, is parsed. In this case, the decoder adds motion vector differential values to which a negative sign or a positive sign is applied to each of a plurality of MVP candidates (MVP candidate A, MVP candidate B) to obtain a plurality of candidate reference blocks (Candidate reference block). A-1, Candidate reference block A-2, Candidate reference block B-1, Candidate reference block B-2).

Next, the decoder determines that all of the plurality of reference block candidates (Candidate reference block A-1, Candidate reference block B-1) based on a particular component of the motion vector difference value to which the negative sign is applied are outside the boundary of the reference picture. It is possible to determine whether it is located in If all of the plurality of reference block candidates based on a particular component of the motion vector difference value to which the negative sign is applied are located outside the boundary of the reference picture, the decoder may select the current block without parsing the code information of the particular component. can determine the sign of a particular component of the motion vector difference value. In this case, the decoder can use the positive sign as the sign of the specific component of the motion vector difference value of the current block.

In addition, the decoder detects that all of the plurality of reference block candidates (Candidate reference block A-2, Candidate reference block B-2) based on a specific component of the motion vector difference value to which the positive sign is applied are outside the boundary of the reference picture. You can also determine where it is located. If all of a plurality of reference block candidates based on a particular component of a motion vector difference value to which a positive sign is applied are located outside the boundary of the reference picture, the decoder may select the current block without parsing the code information of the particular component. can determine the sign of a particular component of the motion vector difference value. In this case, the decoder can use the negative sign as the sign of the specific component of the motion vector difference value of the current block.

FIG. 27 illustrates how the sign bit of the motion vector difference value of the current block is implicitly signaled according to an embodiment of the present invention. According to one embodiment, the resolution of the motion vector difference value of the current block may not be determined before the sign of the motion vector difference value of the current block is determined. For example, the sign of the motion vector difference value of the current block can be determined before parsing the resolution indicator indicating the resolution of the motion vector difference value of the current block on the syntax for video signal processing. In this case, the decoder multiplies any one of a plurality of available resolutions (res 1, res 2) to present a motion vector difference value to which either one of a negative sign and a positive sign is applied. A plurality of reference block candidates (Candidate reference block 1-1, Candidate reference block 1-2, Candidate reference block 2-1, Candidate reference block 2-2) can be obtained by adding to the block motion vector predictor.

Next, the decoder determines that all of the plurality of reference block candidates (Candidate reference block 2-1, Candidate reference block 1-1) based on a particular component of the motion vector difference value to which the negative sign is applied are outside the boundary of the reference picture. It is possible to determine whether it is located in At this time, the available resolutions applied to the motion vector difference values to construct each of the plurality of reference block candidates based on the specific components of the motion vector difference values to which the negative sign is applied may be different. If all of a plurality of reference block candidates based on a particular component of a motion vector difference value to which a negative sign is applied are located outside the boundary of the reference picture, the decoder may select the current block without parsing the code information of the particular component. can determine the sign of a particular component of the motion vector difference value. In this case, the decoder can use the positive sign as the sign of the specific component of the motion vector difference value of the current block.

Also, the decoder determines that all of the plurality of reference block candidates (Candidate reference block 1-2, Candidate reference block 2-2) based on a specific component of the motion vector difference value to which the positive sign is applied are outside the boundary of the reference picture. You can also determine where it is located. At this time, the available resolutions applied to the motion vector difference values to construct each of the plurality of reference block candidates based on the specific components of the motion vector difference values to which the positive sign is applied may differ from each other. If all of a plurality of reference block candidates based on a particular component of a motion vector difference value to which a positive sign is applied are located outside the boundary of the reference picture, the decoder may select the current block without parsing the code information of the particular component. can determine the sign of a particular component of the motion vector difference value. In this case, the decoder can use the negative sign as the sign of the specific component of the motion vector difference value of the current block.

25 to 27 exemplify the x-axis component of the motion vector difference value for convenience of explanation, but the present disclosure is not limited thereto. For example, the above-described embodiments can be applied in the same or corresponding manner to the y-axis component of motion vector difference values.

FIG. 28 shows an example syntax for the embodiment of FIGS. 25-27. Referring to the first syntax order 2801 of FIG. 28, the decoder determines whether to parse the sign information of the motion vector difference value based on the absolute value (|MVD|) of the motion vector difference value. be able to. Referring to the second syntax order 2802 of FIG. 28, the decoder parses the sign information of the motion vector difference value based on the absolute value (|MVD|) of the motion vector difference value and the resolution (MVD resolution) of the motion vector difference value. You can decide whether to Referring to the third syntax order 2803 of FIG. 28, the decoder determines the motion vector based on the absolute value of the motion vector difference value (|MVD|), the resolution of the motion vector difference value (MVD resolution), and the MVP flag (MVP flag). It can be determined whether to parse the sign information of the difference value.

FIG. 29 is a diagram illustrating how the resolution and MVP of the motion vector difference value of the current block are signaled according to an embodiment of the present invention. According to one embodiment of the invention, more than one information associated with a motion vector can be signaled based on one integration indicator. For example, the resolution of the motion vector difference value of the current block and the MVP of the current block can be signaled by one integration indicator (MVR Index).

As shown in FIG. 29, the available resolutions mapped to each MVP index can be preset. The encoder and decoder can share a table that maps each of the MVP indices and the available resolutions to each other. In addition, the encoder and decoder can obtain the MVP of the current block and the resolution of the motion vector differential value of the current block based on the shared table and integration index (MVR Index).

Meanwhile, the accuracy of the motion vector may vary depending on the resolution used to modify the motion vector difference value. For example, the greater the resolution used to modify the motion vector difference value, the less accurate the motion vector may be. Accordingly, additional motion vector differential values of the current block can be signaled when a specific available resolution is used for the resolution of the motion vector differential values of the current block. Hereinafter, the additional motion vector difference values may be referred to as second to n-th motion vector difference values. Also, the existing motion vector difference value excluding the additional motion vector difference value may be referred to as a first motion vector difference value.

FIG. 30 illustrates a method of deriving a motion vector of a current block based on multiple motion vector difference values according to an embodiment of the present invention. According to an embodiment of the present invention, a motion vector of a current block can be obtained using a plurality of motion vector difference values and resolutions corresponding to each. For example, if the resolution of the motion vector difference value of the current block is greater than the preset resolution, additional motion vector difference values can be signaled. According to a specific embodiment, the current block's resolution set may include available resolutions of 1/4, 1, and 4 samples. At this time, if the resolution of the motion vector difference value of the current block is '4', an additional motion vector difference value can be signaled. Also, the additional motion vector difference values can be signaled based on a resolution smaller than the resolution at which the existing motion vector difference values were signaled. Through this, the encoder and decoder can improve motion vector accuracy and reduce prediction errors.

Referring to FIG. 30, R1 indicates the resolution of the first motion vector difference value, and R2 indicates the resolution of the second motion vector difference value. MVDval1 indicates the first motion vector difference value signaled in units of R1, and MVDval2 indicates the second motion vector difference value signaled in units of R2. For example, if the resolution of the first motion vector difference value (R1) is greater than the preset resolution, the decoder can additionally acquire the signaled second motion vector difference value (MVDval2). At this time, the resolution (R2) of the second motion vector difference value may be smaller than the resolution (R1) of the first motion vector difference value of the current block.

Meanwhile, an inter-prediction method or motion compensation method for the current block according to an embodiment of the present invention may include affine model-based motion compensation (hereinafter referred to as affine motion compensation). According to the existing inter prediction method, inter prediction is performed using only one motion vector for L0 prediction and L1 prediction for the current block. This makes existing general inter-prediction methods optimized for prediction of translational motion. However, in order to efficiently perform motion compensation for zoom in/out, rotation, and other irregular motions, reference blocks of various shapes and sizes should be used. The following describes how affine motion compensation is performed according to an embodiment of the present invention.

FIG. 31 is a diagram illustrating motion compensation based on an affine model according to one embodiment of the present invention. Referring to FIG. 31, in affine motion compensation, a current block 3101 can be predicted using a reference block 3102 having a size, shape and/or direction different from that of the current block 3101 . That is, reference block 3102 may have a non-rectangular shape and may be larger or smaller in size than current block 3101 . A reference block 3102 can be obtained by performing an affine transformation on the current block 3101 . Scaling, rotation, shearing, reflection, or orthogonal projection for the current block can be performed through affine transformation. According to one embodiment, the affine transformation can be performed using multiple Control Point Motion Vectors (CPMV). For example, the affine transforms can include a 6-parameter affine transform with three control point motion vectors and a 4-parameter affine transform with two control point motion vectors. A specific embodiment for this will be described later.

FIG. 32 illustrates one embodiment of a 4-parameter affine motion compensation method. Affine motion compensation can be performed using a preset number of control point motion vectors (CPMV) in order to reduce the computational complexity and signaling overhead of the affine transformation. A control point motion vector (CPMV) is a motion vector corresponding to a specific control point (or sample position) of the current block. The specific control point may include at least one corner of the current block. In an embodiment of the present invention, the CPMV corresponding to the upper left corner of the current block is v0 (or the first CPMV), and the CPMV corresponding to the upper right corner of the current block is v1 (or the second CPMV). , the CPMV corresponding to the lower left corner of the current block is called v2 (or the third CPMV). A CPMV set containing at least two CPMVs can be used for affine motion compensation.

According to the embodiment of FIG. 32, 4-parameter affine motion compensation can be performed using v0 and v1. A current block 3201 indicated by a solid line can be predicted using a reference block 3202 indicated by a dotted line. Each sample of the current block 3201 can be mapped to different reference samples through affine transformation. More specifically, the motion vector (v _x , v _y ) at the sample position (x, y) of the current block 3201 can be derived from Equation 2 below. In this disclosure, the sample position may indicate relative coordinates within the current block. For example, the sample position (x, y) may be coordinates with the position of the upper left sample of the current block as the origin (0, 0).

where (v _0x , v _0y ) is the first CPMV corresponding to the upper left corner of the current block 3201, and (v _1x , v _1y ) is the second CPMV corresponding to the upper right corner of the current block. be. Also, w is the width of the current block 3201 .

FIG. 33 illustrates one embodiment of a 6-parameter affine motion compensation method. Affine motion compensation using three or more CPMVs can be performed for accurate prediction of more complex motion. Referring to FIG. 33, 6-parameter affine motion compensation can be performed using three CPMVs, namely v0, v1, and v2. Here, v0 is the CPMV corresponding to the upper left corner of the current block, v1 is the CPMV corresponding to the upper right corner of the current block, and v2 is the CPMV corresponding to the lower left corner of the current block. A motion vector of each sub-block of the current block can be calculated based on the v0, v1 and v2. According to the embodiment of FIG. 33, the current block 3301 indicated by the solid line can be predicted using the reference block 3302 at the position indicated by the dotted line. Each sample of the current block 3301 can be mapped to different reference samples through affine transformation. More specifically, the motion vector (mv ^x , mv ^y ) at the sample position (x, y) of the current block 3301 can be derived from Equation 3 below.

where (mv ₀ ^x , mv ₀ ^y ) is the first CPMV corresponding to the upper left corner of the current block 3301 and (mv ₁ ^x , mv ₁ ^y ) is the second CPMV corresponding to the upper right corner. and (mv ₂ ^x , mv ₂ ^y ) is the third CPMV corresponding to the lower left corner. Also, w is the width of the current block 3301 and h is the height of the current block 3301 .

FIG. 34 illustrates an embodiment of a sub-block-based affine motion compensation method. Using affine motion transformation, a motion vector (ie, motion vector field) at each sample position of the current block can be derived. However, in order to reduce the amount of computation, sub-block-based affine motion compensation can be performed as shown in FIG. Referring to FIG. 34, the current block can contain multiple sub-blocks. Also, a representative motion vector for each sub-block can be obtained based on the CPMV set (v0, v1, and v2). According to one embodiment, the representative motion vector of each sub-block may be a motion vector corresponding to the middle sample position of the corresponding sub-block. According to additional embodiments, a motion vector with higher accuracy than a general motion vector can be used as the motion vector of the sub-block. For this, a motion compensated interpolation filter can be applied.

The size of sub-blocks on which affine motion compensation is performed can be set in various ways. According to an embodiment of the present invention, sub-blocks may have preset sizes such as 4×4 or 8×8. According to another embodiment of the present invention, the sub-block size M×N can be determined by Equation 4 below.

where w is the width of the current block and MvPre is the fractional accuracy of the motion vector. ( _v2x , _v2y ) is the third CPMV corresponding to the lower left corner of the current block. According to one embodiment, the third CPMV can be calculated according to Equation 2 above. max(a, b) is a function that returns the larger value of a and b, and abs(x) is a function that returns the absolute value of x. Also, clip3(x, y, z) is a function that returns x if z<x, returns y if z>y, and returns z otherwise.

The decoder obtains motion vectors for each sub-block of the current block using the CPMVs of the CPMV set. Also, the decoder can obtain a predictor of each sub-block of the current block using the representative motion vector of each sub-block. A predictor of the current block can be obtained by combining predictors of each sub-block, and a decoder can reconstruct the current block using the predictors of the current block thus obtained.

Figures 35, 36, 37, and 38 illustrate embodiments of methods for obtaining control point motion vector sets for prediction of the current block. According to an embodiment of the present invention, a CPMV set for affine motion compensation of the current block can be obtained through various methods. More specifically, a CPMV set for prediction of the current block can be obtained by referring to motion vector information sets of one or more neighboring blocks. Also, a motion vector information set indicates a set of motion vector information of one or more blocks. A peripheral block may indicate a block including a preset peripheral position of the current block. At this time, the peripheral block may be a coding unit including a preset peripheral position or a preset unit (eg, 4×4, 8×8) region including the peripheral position.

There can be multiple candidates that can be referenced to derive the CPMV of the current block. Therefore, information about neighboring blocks referred to in order to derive the CPMV of the current block can be separately signaled. According to an embodiment of the present invention, a CPMV indicator can be signaled that indicates a motion vector information set to refer to derive a motion vector of each sub-block of the current block. The encoder can signal the CPMV indicator. The CPMV indicator can indicate the motion vector information set of neighboring blocks to be referred to derive the motion vector of each sub-block of the current block. The decoder can acquire the indicator and refer to the motion vector information sets of the neighboring blocks indicated by the indicator to acquire each CPMV of the CPMV set for the current block.

According to a more specific embodiment, the encoder and decoder can generate a CPMV candidate list consisting of one or more motion vector information set candidates. Each motion vector information set candidate constituting the CPMV candidate list is a motion vector set of neighboring blocks that can be used to derive the motion vector of the current block. At this time, the CPMV indicator may be an index indicating any one motion vector information set in the CPMV candidate list. The CPMV of the current block can be obtained by referring to the motion vector information set selected based on the CPMV indicator (ie, index) from the CPMV candidate list. Various embodiments of motion vector information set candidates that can be included in a CPMV candidate list for deriving motion vector information (or CPMV set) of the current block will be described below.

FIG. 35 illustrates one embodiment of obtaining the CPMV set for the current block. The embodiment of FIG. 35 assumes that the CPMV set of the current block contains two CPMVs, namely v0 and v1. According to an embodiment of the present invention, the CPMV of the current block can be derived from motion vectors of neighboring blocks adjacent to the corresponding point. Referring to FIG. 35, v0 can be derived from the motion vector of any one of neighboring blocks A, B, and C adjacent to the corresponding point, and v1 can be derived from neighboring blocks D and E adjacent to the corresponding point. can be derived from any one motion vector. When the motion vectors of neighboring blocks A, B, C, D, and E are vA, vB, vC, vD, and vE, respectively, the motion vector information sets that can be included in the CPMV candidate list are given by the following [formula 5] can be derived.

That is, a (v0, v1) pair consisting of v0 selected from vA, vB, and vC and v1 selected from vD and vE can be obtained. At this time, v0 is derived from the motion vector of the block adjacent to the upper left corner of the current block, and v1 is derived from the motion vector of the block adjacent to the upper right corner of the current block. According to additional embodiments, motion vector scaling may be performed based on a POC (Picture Order Count) of the current block, POCs of reference pictures of neighboring blocks, and POCs of reference pictures of the current block.

A CPMV candidate list including motion vector information set candidates obtained in this way may be generated, and a CPMV indicator indicating any one motion vector information set in the CPMV candidate list may be signaled. According to additional embodiments of the present invention, the CPMV candidate list may also include motion vector information set candidates for inter-prediction of other schemes. For example, the CPMV candidate list may include motion vector information set candidates obtained based on existing MVP candidates for inter-prediction.

A decoder can derive the CPMV of the current block based on the motion vector information set obtained from the candidate list. According to one embodiment, the decoder uses the motion vector of the motion vector information set obtained from the candidate list for the CPMV of the current block without a separate motion vector difference value, thus performing affine merge prediction. be able to. Such affine motion compensation methods can be referred to as affine merge prediction modes.

According to another embodiment, the decoder can obtain a separate motion vector difference value for the CPMV of the current block. The decoder can obtain the CPMV of the current block by summing the motion vector of the motion vector information set obtained from the CPMV candidate list with the motion vector difference value. Such an affine motion compensation method can be called an affine inter prediction mode. A flag or index may be separately signaled to indicate whether the decoder uses a separate motion vector difference value for affine motion compensation of the current block. According to additional embodiments, whether to use a separate motion vector difference value for affine motion compensation of the current block may be determined based on the size of the current block (eg, CU size). For example, if the size of the current block is greater than or equal to a preset size, the encoder and decoder can be set to use separate motion vector difference values for affine motion compensation of the current block.

FIG. 36 illustrates another embodiment of obtaining the CPMV set of the current block. According to another embodiment of the present invention, the CPMV of the current block can be derived from motion vector information of neighboring blocks on which affine motion compensation has been performed. That is, the CPMV of the current block can be derived from the CPMV or motion vector of neighboring blocks. At this time, the peripheral blocks may include a left peripheral block of the current block and an upper peripheral block of the current block. Referring to FIG. 36(a), left peripheral blocks include blocks adjacent to the lower left corner of the current block, ie, left block A and lower left block D. Referring to FIG. The upper peripheral blocks include upper left block E adjacent to the upper left corner of the current block, and upper block B and upper right block C adjacent to the upper right corner of the current block. The decoder checks whether affine motion compensation has been performed on neighboring blocks in a preset order. According to one embodiment, the preset order may be A, B, C, D, E. If a neighboring block on which affine motion compensation is performed is found, the decoder acquires the CPMV set of the current block using the CPMV set (or motion vector) of the corresponding neighboring block. Referring to the embodiment of FIG. 36(b), the CPMV set of left block A can be used to derive the CPMV set of the current block. That is, the CPMV set (v0, v1) of the current block can be obtained based on the CPMV set (v2, v3, v4) of the left block A.

According to an embodiment of the present invention, information on neighboring blocks referred to for deriving the CPMV of the current block can be separately signaled. At this time, each of the CPMV sets of neighboring blocks of the current block can be a motion vector information set candidate constituting the CPMV candidate list according to a preset order. More specifically, the motion vector information set candidate is derived from the first candidate derived from the CPMV (or motion vector) of the left neighboring block of the current block and the CPMV (or motion vector) of the upper neighboring block of the current block. can include a second candidate that has been At this time, the left peripheral block is a block adjacent to the lower left corner of the current block, and the upper peripheral block is a block adjacent to the upper left corner of the current block or a block adjacent to the upper right corner of the current block. A CPMV candidate list including motion vector information set candidates obtained in this way may be generated, and a CPMV indicator indicating any one motion vector information set in the CPMV candidate list may be signaled. According to an embodiment, the CPMV indicator may indicate location information of neighboring blocks referenced to derive motion vectors of each sub-block of the current block. The decoder can obtain the CPMV set of the current block by referring to the CPMV set (or motion vector) of neighboring blocks indicated by the CPMV indicator.

According to an additional embodiment of the present invention, the CPMV of the current block can be derived based on the CPMV of neighboring blocks near the corresponding point. For example, v0 can be obtained by referring to the CPMV of the left peripheral block, and v1 can be obtained by referring to the CPMV of the upper peripheral block. Alternatively, v0 can be obtained by referring to the CPMV of neighboring blocks A, D, or E, and v1 can be obtained by referring to the CPMV of neighboring blocks B or C.

Figure 10 illustrates another embodiment of obtaining the CPMV set of the current block; According to one embodiment, the CPMV of the current block can be derived from the CPMV or motion vector of the block on which affine motion compensation has been performed. For example, when a block corresponding to a preset position based on the current block is subjected to affine motion compensation, the CPMV of the current block can be derived from the CPMV set or at least one motion vector of the corresponding block. Referring to FIG. 37, the preset positions may be A0 and A1 adjacent to the lower left corner of the current block, B0 and B1 adjacent to the upper right corner, and the upper left B2 position. In this embodiment, the preset positions may include positions that are not adjacent to the current block. For example, the preset position may include a position corresponding to the position of the current block within a picture other than the current picture. That is, the CPMV set of the current block can be derived from spatial candidates or temporal candidates corresponding to preset positions. A motion information set candidate obtained by the method described through FIG. 37 can be called an inherent candidate or an affine merge candidate.

As described with reference to FIGS. 36 and 37, the CPMV of the current block can be derived from motion vector information of neighboring blocks on which affine motion compensation has been performed. According to a specific embodiment, a CPMV set ((v_0x, v_0y), (v_1x, v_1y)) for 4-parameter affine motion compensation for the current block can be derived as [Equation 6].

Here, (v_E0x, v_E0y) represent motion vectors used for affine motion compensation of the upper left block of the current block, and (v_E1x, v_E1y) represent motion vectors used for affine motion compensation of the upper right block of the current block. and (v_E2x, v_E2y) may indicate a motion vector used for affine motion compensation of the lower left block of the current block.

FIG. 38 illustrates a method of obtaining control point motion vectors of the current block according to another embodiment of the present invention. According to one embodiment, the CPMV of the current block can be obtained by referring to one or more motion vectors around the current block. At this time, one or more motion vectors around the current block may include motion vectors other than motion vectors used for affine motion compensation. For example, the CPMV of the current block can be derived from the motion vector of the preset position based on the corresponding point. For example, the preset position may be a position included in a block within a certain distance from the point corresponding to the CPMV to be guided.

Referring to FIG. 38, the first CPMV (mv0), the second CPMV (mv1), and the third CPMV (mv2) for affine motion compensation of the current block are the motions of the previously set positions. It can be derived from vectors. For example, the first CPMV (mv0) can be derived from motion vector information for each of locations A, B, C around the upper left corner. A second CPMV (mv1) can be derived from motion vector information for each of positions D, E around the upper right corner. A third CPMV (mv2) can be derived from the motion vector information for each of the positions F, G around the lower left corner.

According to an embodiment, the decoder can check the availability of the motion vectors of the CPMV peripheral positions in a preset order. If a usable motion vector is found, the decoder can obtain the CPMV using the corresponding motion vector. Also, a preset combination of peripheral positions referred to for deriving the CPMV for each point of the current block can be used. A motion information set candidate obtained by the method described through FIG. 38 can be called a constructed candidate or an affine inter candidate. The motion vector information sets illustrated in FIGS. 35 to 38 can be motion vector information set candidates forming the CPMV candidate list according to a preset order.

FIG. 39 illustrates a method of obtaining control point motion vectors of the current block according to another embodiment of the present invention. As described above, multiple CPMVs can be used for affine motion compensation of the current block. According to an embodiment of the present invention, some of the plurality of CPMVs included in the CPMV set for motion compensation of the current block can be derived based on others. For example, the encoder and decoder can obtain the CPMV corresponding to each of the two points of the current block by the method described above. Then, the encoder and decoder can derive CPMVs corresponding to different points of the current block from the previously acquired CPMVs.

Referring to FIG. 39, the first CPMV (mv ₀ ^x , mv ₀ ^y ) and the second CPMV (mv ₁ ^x , mv ₁ ^y ) are derived from the third CPMV (mv ₂ ^x , mv ₂ ^y ). Alternatively, the second CPMV (mv ₁ ^x , mv ₁ ^y ) can be derived from the first CPMV (mv ₀ ^x , mv ₀ ^y ) and the third CPMV (mv ₂ ^x , mv ₂ ^y ). In the formula of FIG. 39, w and h can be the width and height of the current block, respectively.

FIG. 40 illustrates how the control point motion vector of the current block is obtained according to one embodiment of the present invention. As described above, the CPMV set of the current block can consist of multiple CPMVs. Also, it can be obtained based on a CPMV predictor and a motion vector difference value for each of a plurality of CPMVs. The CPMV motion vector difference value can be signaled from the encoder to the decoder. That is, a motion vector difference value used to derive a CPMV for each CPMV for affine motion compensation of the current block can be signaled. Hereinafter, the CPMV motion vector difference value may be referred to as the CPMV difference value. In FIG. 40, mv0, mv1, and mv2 displayed in the upper bar are CPMV predictions of the first CPMV (mv0), the second CPMV (mv1), and the third CPMV (mv2), respectively. child can be shown.

FIG. 40(a) shows how the first CPMV (mv0) and the second CPMV (mv1) of the current block are obtained when 4-parameter affine motion compensation is performed. A first CPMV difference value (mvd0) and a second CPMV difference value (mvd1) for each of the first CPMV (mv0) and the second CPMV (mv1) can be signaled, respectively.

FIG. 40(b) shows that the first CPMV (mv0), the second CPMV (mv1), and the third CPMV (mv2) of the current block are obtained when 6-parameter affine motion compensation is performed. Show how. A first CPMV difference value (mvd0), a second CPMV difference value (mvd1), and a third of CPMV difference values (mvd2) can be signaled respectively.

FIG. 41 illustrates how the current block's control point motion vector difference value is signaled according to one embodiment of the present invention. The CPMV difference value of the current block can be signaled according to the syntax described above with reference to FIG. Also, the decoder can obtain a CPMV difference value (IMVD) of the current block based on at least one information about the signaled motion vector difference value. In FIG. 41, compldx is an index indicating a component of the motion vector difference value, and may be 0 or 1. For example, compldx can indicate either the x-axis component or the y-axis component of the motion vector difference value. Also, the CPMV difference value of the current block may be different for each reference picture list. In FIG. 41, L0 indicates the first reference picture list and L1 indicates the second reference picture list.

Meanwhile, a CPMV difference value corresponding to a specific point of the current block may be similar to CPMV difference values corresponding to different points of the current block. Accordingly, a CPMV difference value corresponding to a specific point of the current block can be obtained based on CPMV difference values corresponding to different points of the current block. Through this, the encoder and decoder can reduce signaling overhead for the CPMV difference value.

FIG. 42 illustrates a method of obtaining control point motion vectors of the current block according to another embodiment of the present invention. Referring to FIG. 42, a CPMV set for affine motion compensation of the current block can be derived based on at least one common CPMV difference value. For example, a CPMV corresponding to a specific point of the current block can be obtained based on CPMV difference values corresponding to different points of the current block.

The encoder can signal one common CPMV difference value and at least one additional difference value for affine motion compensation of the current block. Also, the decoder can obtain a CPMV set for affine motion compensation of the block using one common CPMV difference value and at least one additional difference value. In FIG. 42, mv0, mv1, and mv2 displayed in the upper bar are CPMV predictions of the first CPMV (mv0), the second CPMV (mv1), and the third CPMV (mv2), respectively. child can be shown.

FIG. 42(a) shows how the first CPMV (mv0) and the second CPMV (mv1) of the current block are obtained when 4-parameter affine motion compensation is performed. A first CPMV difference value (mvd0) for the first CPMV (mv0) can be signaled. Also, the first additional difference value (mvd1') used to obtain the second CPMV difference value can be signaled. Specifically, the second CPMV difference value can be expressed by adding the first additional difference value (mvd1') to the first CPMV difference value (mvd0). That is, the second CPMV (mv1) can be obtained based on the second CPMV predictor, the first CPMV difference value (mvd0), and the first additional difference value (mvd1'). If the first CPMV difference value and the second CPMV difference value are similar, the first additional difference value (mvd1') may be less than the second CPMV difference value. Accordingly, the encoder and decoder can reduce signaling overhead for the CPMV difference value compared to the method described through FIG.

FIG. 42(b) shows that the first CPMV (mv0), the second CPMV (mv1), and the third CPMV (mv2) of the current block are obtained when 6-parameter affine motion compensation is performed. Show how. A first CPMV difference value (mvd0) for the first CPMV (mv0) can be signaled. Also, a first additional difference value (mvd1') and a second additional difference value (mvd2') used to obtain the second CPMV difference value and the third CPMV difference value, respectively, can be signaled. Specifically, the second CPMV difference value can be expressed by adding the first additional difference value (mvd1') to the first CPMV difference value (mvd0). Also, the third CPMV difference value can be expressed by adding the second additional difference value (mvd2') to the first CPMV difference value (mvd0).

FIG. 43 shows how the control point motion vector difference value is signaled when the control point motion vector of the current block is obtained according to the embodiment described through FIG. As described above, the CPMV set for affine motion compensation of the current block can include multiple CPMVs. At this time, some CPMV difference values (mvdLK) among the plurality of CPMVs can be obtained based on the difference values of other CPMVs and additional difference values. In FIG. 43, lmvd may be any one of a preset CPMV difference value or at least one additional difference value.

In FIG. 43, cpIdx can indicate the control point index of the current block. For example, cpIdx may be 0 or 1 when 4-parameter affine motion compensation is performed on the current block. cpIdx can be 0, 1, or 2 when 6-parameter affine motion compensation is performed for the current block. According to one embodiment, the CPMV difference value (MvdLx) can be determined in different ways depending on cpIdx. For example, if cpIdx is a preset value (eg, '0'), MvdLx may be a preset CPMV difference value (lMvd[0][compIdx]). If cpIdx is not a preset value (e.g. '0'), MvdLx is a value obtained by adding an additional differential value corresponding to the corresponding index to the preset CPMV difference value (lMvd[0][compIdx]). Possible. where MvdLx can denote the difference between CPMV and the CPMV predictor. That is, MvdLx can be (CPMV-CPMV predictor).

Also, compIdx is an index indicating a component of the motion vector difference value, and may be 0 or 1. For example, compIdx can indicate either the x-axis component or the y-axis component of the motion vector difference value. Also, the CPMV difference value of the current block may be different for each reference picture list. In FIG. 41, L0 indicates the first reference picture list and L1 indicates the second reference picture list.

FIG. 44 is a diagram illustrating how the control point motion vector difference value of the current block is signaled according to one embodiment of the present invention. Referring to FIG. 44, the CPMV difference value can be encoded by a method similar to the signaling method of the motion vector difference value described above with reference to FIG. For example, the encoder can generate at least one piece of information about the CPMV difference value by encoding the CPMV difference value separately according to the control point index (cpIdx). Also, the encoder can signal at least one piece of information for the encoded CPMV difference value. A decoder can obtain a CPMV difference value for each control point index (cpIdx) based on at least one piece of information about the CPMV difference value.

Embodiments related to the resolution of the motion vector difference values described above can also be applied to the CPMV difference values described through the embodiments of FIGS. 9, 43, and 44 in the same or corresponding manner. For example, if the resolution of the CPMV difference values of the current block is R, then the signaled CPMV difference values can be modified based on R. At this time, the decoder can obtain a corrected CPMV difference value by multiplying the signaled CPMV difference value by the resolution (R).

FIG. 45 is a diagram illustrating how a motion vector is obtained using a differential predictor for a current block control point motion vector differential value according to an embodiment of the present invention. Referring to FIG. 45, a CPMV difference value indicating the difference between the CPMV of the current block and the CPMV predictor can be determined based on the differential predictor (mvdp). Here, the difference predictor (mvdp) can indicate the predicted value for the difference between CPMV and the CPMV predictor. Specifically, a CPMV difference value can be obtained based on a difference predictor (mvdp) and additional difference values (mvd0″, mvd1″, mvd2″).

For example, a differential predictor (mvdp) and additional differential values (mvd0″, mvd1″, mvd2″) can be signaled from the encoder to the decoder. ) can be encoded or decoded by the method described above with reference to FIG. 9 or FIG. In FIG. 45, mv0, mv1, and mv2 displayed in the upper bar are CPMV predictions of the first CPMV (mv0), the second CPMV (mv1), and the third CPMV (mv2), respectively. child can be shown. Encoders and decoders can use differential predictors (mvdp) to reduce signaling overhead for CPMV differential values.

Meanwhile, according to an embodiment of the present invention, a differential predictor can be obtained based on at least one of a plurality of CPMVs included in the CPMV set of the current block. For example, the difference predictor may be any one CPMV difference value among a plurality of CPMVs included in the CPMV set of the current block. The following describes how the differential predictors used to derive the CPMV set of the current block are obtained.

FIG. 46 illustrates how the control point motion vector of the current block is obtained using the differential predictor according to one embodiment of the present invention. According to an embodiment of the present invention, the encoder and decoder can obtain a differential predictor (mvdp) of the current block using at least one of mvd0, mvd1, and mvd2 (hereinafter mvdx) of the current block. . According to one embodiment, the differential predictor (mvdp) can be set to a value of mvdx that is less than the lowest absolute value of mvdx. Alternatively, the differential predictor (mvdp) can be set to a value greater than the absolute value of mvdx with the largest absolute value among mvdx. In this case, the signs of additional difference values (mvd0″, mvd1″, mvd2″ in FIG. 45) used to derive each of the plurality of CPMVs included in the CPMV set of the current block can match. The difference value can indicate the difference between mvdx and the difference predictor (mvdp).

For example, when the differential predictor (mvdp) is set to a value smaller than the absolute value of mvdx with the smallest absolute value among mvdx, the sign of the additional differential values (mvd0″, mvd1″, mvd2″ in FIG. 45) is The positive sign (+) can be matched, and if the difference predictor (mvdp) is set to a value greater than the absolute value of mvdx with the largest absolute value among mvdx, the additional difference value (mvd0 ", mvd1", mvd2") can match the negative sign (-).

Referring to FIG. 46, the differential predictor (mvdp) can be set to a value smaller than the absolute value of mvdx with the smallest absolute value among mvdx. In FIG. 46, the horizontal dashed line indicates the smallest absolute value of each x-axis component of mvdx, and the vertical dashed line indicates the smallest absolute value of each y-axis component of mvdx. The right side of the x-axis can have a larger value, and the higher side of the y-axis can have a larger value. At this time, the differential predictor (mvdp) may be a value within a third quadrant (mvdp area) among quadrants partitioned by the fracture line.

In this case, the signs of the additional difference values of the current block can all be positive signs (+). Accordingly, information indicating a sign for the additional difference value may not be separately signaled. Specifically, each of the difference predictor (mvdp) and the additional difference value (mvdx″) of the current block can be derived as [Equation 7].

Referring to [Formula 7], information (mvd_sign_flag) indicating the sign of the differential predictor (mvdp) can be signaled. The encoder can signal the sign information of the differential predictor. Also, the decoder can obtain the CPMV set of the current block based on the sign information (mvd_sign_flag) of the differential predictor. On the other hand, information indicating the sign of each of the additional difference values (mvdx″) may not be signaled. According to an additional embodiment, an mvdp indicator indicating how to determine the difference predictor (mvdp) is signaled. For example, the mvdp indicator can be a flag indicating whether the differential predictor (mvdp) is determined to be less than the minimum value or greater than the maximum value of mvdx.

FIG. 47 is a diagram illustrating how a control point motion vector of a current block is obtained using a differential predictor according to another embodiment of the present invention. According to an embodiment of the present invention, the differential predictor of the current block may be any one CPMV differential value among a plurality of CPMVs included in the CPMV set of the current block. Referring to FIG. 46, a difference predictor can be changed by Method 0, Method 1, and Method 2. FIG. For example, in Method 0 the differential predictor can be mvd0. Also, in Method 1, the differential predictor may be mvd1. In Method 2, the differential predictor can be mvd2.

According to one embodiment, an mvdp indicator (cpIdxPred) can be signaled that indicates which of multiple methods the differential predictor is determined by. Here, the plurality of methods may include at least one of Method 0, Method 1, and Method 2 of FIG. 47 and the methods described with reference to FIG. In the FIG. 47 embodiment, the encoder can signal an mvdp indicator (cpIdxPred) that indicates one of Method 0, Method 1, Method 2. Also, the decoder can determine one of Method 0, Method 1, and Method 2 based on the mvdp indicator (cpIdxPred). Also, the decoder can obtain a differential predictor of the current block according to the determined method. Also, the decoder can obtain a CPMV set for the current block based on the obtained differential predictors.

According to another embodiment, the mvdp indicator (cpIdxPred) can also be implicitly signaled. That is, the encoder and decoder can also determine the differential predictor of the current block without signaling the mvdp indicator (cpIdxPred). This will be described in detail with reference to FIGS. 49 to 50. FIG.

FIG. 48 illustrates how differential predictors are obtained according to one embodiment of the present invention. According to an embodiment of the present invention, if the control point index (cpIdx) is the same as the mvdp indicator (cpIdxPred), the differential predictor can be used as the differential value of the corresponding CPMV. If the control point index (cpIdx) is not the same as the mvdp indicator (cpIdxPred), the difference value of the corresponding CPMV can be obtained based on the additional difference value and the difference predictor corresponding to cpIdx. An mvdp indicator (cpIdxPred) can be signaled to indicate a coding efficient method for motion vector differential values. In FIG. 48, lmvd can indicate either one of the difference predictor or the additional difference value. Also, MvdLK can indicate the CPMV difference value.

In FIG. 48, LX can indicate the reference picture list L0 or L1. Also, the mvdp indicator can be expressed in bits of variable length. For example, the mvdp indicator can be signaled by a truncated unary method. Also, the embodiments described with reference to FIGS. 46 to 48 can be performed for each component of the motion vector difference value. Embodiments related to the resolution of the motion vector difference value described above can also be applied to the CPMV difference value described through the embodiments of FIGS. 46 to 48 in the same or corresponding manner. For example, if the resolution of the CPMV difference values of the current block is R, then the signaled CPMV difference values can be modified based on R. At this time, the decoder can obtain a corrected CPMV difference value by multiplying the signaled CPMV difference value by the resolution (R).

According to one embodiment of the present invention, the mvdp indicator (cpIdxPred) can be signaled or determined in a manner that reduces IMvd signaling for control point indices that are not mvdp indicator (cpIdxPred). For example, the mvdp indicator can be signaled to match the sign of the IMvd for control point indices that are not mvdp indicators. Alternatively, differential predictors can be determined without explicit signaling. This will be described later with reference to FIGS. 49 to 51. FIG.

As described above, if the difference predictor of the current block is determined to be the minimum or maximum absolute value of the CPMV difference values of the current block, the sign between lmvd for the control point index can match. At this time, if the control point index is different from the mvdp indicator, the code information may not be signaled. Here, the minimum or maximum absolute value of the CPMV difference value can mean that the absolute value of the specific component is minimum or maximum. For example, if the mvdp indicator indicates one of three values, one of the mvdp indicators can be signaled using one bit and two using two bits according to the cutting unary method.

According to another embodiment of the present invention, the mvdp indicator may be a CPMV difference value corresponding to a median or intermediate value among absolute values of CPMV difference values. In this case, the sign of lmvd can be one (+) and one (-). This allows one bit to be used to signal sign information for each of the two CPMV difference values. Through this, the encoder and decoder can reduce signaling overhead for sign information of CPMV difference values.

According to yet another embodiment of the present invention, an mvdp indicator (cpIdxPred) that indicates how to determine the differential predictor of the current block can be implicitly signaled. For example, how to determine the differential predictor for the current block can be determined based on other information. More specifically, the encoder and decoder may determine a CPMV difference value having the largest absolute value among the CPMV difference values of the current block as a difference predictor. This is because when the motion vector difference value is signaled, the larger the absolute value of the difference predictor used to derive multiple CPMVs, the more advantageous it may be.

FIG. 49 illustrates how the current block's difference vector predictor is determined according to one embodiment of the present invention. Referring to FIG. 49, three CPMVs can be used for affine motion compensation of the current block. Although the x-axis component will be described below as an example for convenience of explanation, the present disclosure is not limited thereto. The embodiments described below can also be applied to the y-axis component in the same or corresponding manner. According to an embodiment, the control point index indicating the CPMV having the smallest absolute value of the x-axis component among the CPMV difference values mvd0, mvd1, and mvd2 of each CPMV of the current block may be '1'. .

In FIG. 49, the dashed line indicates the value of the x-axis component of the reference predictor. As shown in FIG. 49, when the CPMV difference value with the smallest absolute value of the CPMV difference value is used as the difference predictor, the sign of each of the remaining CPMV difference values that are not selected as the difference predictor is a positive sign (+). can match This allows the encoder to omit signaling for the mvdp indicator. Also, the decoder can derive the CPMV set of the current block without the mvdp indicator.

FIG. 50 is a diagram showing how the control point motion vector difference value of the current block is signaled. As described above with reference to FIGS. 48 and 49, when using the code bit hiding method, signaling of code information when the control point index and the mvdp indicator are different can be omitted. Referring to FIG. 50, sign information (mvd_sign_flag) of the motion vector difference value can be parsed when the control point index and the mvdp indicator are the same. Conversely, if the control point index and the mvdp indicator are different, it may not be performed. In FIG. 50, [0] and [1], where 0 and 1 indicate component indices, can indicate the x-axis component and the y-axis component, respectively.

According to an embodiment of the present invention, information on motion vector difference values can be sequentially parsed according to a preset order. At this time, the preset order may be the order set based on the control point index. According to one embodiment, the preset order can be the order determined based on the mvdp indicator.

According to one embodiment of the present invention, there can be a function for coding motion vector difference values. Also, an example of a function for coding motion vector difference values can be the mvd_coding syntax of FIG. According to one embodiment, a decoder can parse information for motion vector difference values according to control point indices. For example, the mvd_coding syntax of FIG. 50 can be performed for each control point index. At this time, the preset order can be performed as in [Equation 8] below. According to [Formula 8], parsing can always be performed in the same order.

According to another embodiment, information for CPMV difference values corresponding to mvdp indicators can be parsed before information for other CPMV difference values that do not correspond to mvdp indicators. In this case, after decoding the additional difference value corresponding to the specific control point index, the CPMV difference value corresponding to the corresponding control point index can be calculated immediately. Unlike the present embodiment, the information on the CPMV difference value corresponding to the mvdp indicator may not be parsed before the information on the difference value of other CPMVs not corresponding to the mvdp indicator. In this case, even after decoding the additional difference value corresponding to the specific control point index, a waiting time may be required until the CPMV difference value corresponding to the corresponding control point index is calculated. As a specific example, when the mvdp indicator is '1', the mvd_coding syntax is as [Formula 9] below.

As another specific example, when the mvdp indicator is '2', the mvd_coding syntax is as in [Formula 10] below.

FIG. 51 is a diagram illustrating how the control point motion vector of the current block is derived according to another embodiment of the present invention. In FIG. 51, lmvd can indicate the difference predictor or the additional difference value. Referring to FIG. 51, if the specific control point index is the same as the mvdp indicator, a differential predictor can be calculated based on the code information. Alternatively, if the specific control point index is different from the mvdp indicator, sign information may not be used in calculating the additional difference value.

Meanwhile, according to an embodiment of the present invention, the resolution of the motion vector difference value may be different for each of a plurality of CPMVs for motion compensation of the current block. For example, the resolution at which the CPMV difference value is signaled may be different for each control point index. According to an embodiment, when differential predictors are used as in FIGS. 42, 45, and 47, the resolution of the motion vector differential value may be different for each of the CPMVs of the current block.

According to one embodiment, a CPMV difference value of a specific control point index can be used as a difference predictor of the current block. In this case, the resolution for the CPMV difference value corresponding to the corresponding control point index can be signaled. Also, the CPMV difference value corresponding to the corresponding control point index can be signaled based on the corresponding resolution. Meanwhile, CPMV difference values corresponding to control point indexes other than the corresponding control point index can be signaled based on the preset resolution. At this time, the resolution indicating the unit in which the CPMV difference values corresponding to the remaining control point indexes excluding the corresponding control point index are signaled may not be signaled. Also, the preset resolution may be a default value.

According to another embodiment, the CPMV difference value used for the difference predictor of the current block can be signaled based on a relatively small unit of resolution. Also, CPMV difference values of control points other than the CPMV difference value used for the difference predictor of the current block may be signaled based on a relatively large unit resolution.

FIG. 52 illustrates how multiple control point motion vectors are derived for affine motion compensation of the current block according to one embodiment of the present invention. Referring to FIG. 52, a first CPMV difference value (mvd0) and a second CPMV difference value (mvd1) can be obtained based on a common difference predictor (mvdp) and a single absolute difference value (mvdd). can. For example, a first CPMV difference value (Mvd0) can be obtained based on the sum of mvdp and mvdd. Also, a second CPMV difference value (mvd1) can be obtained based on the sum of mvdp and -mvdd.

FIG. 52 illustrates how multiple control point motion vectors are derived for affine motion compensation of the current block according to one embodiment of the present invention. Referring to FIG. 52, a first CPMV difference value (mvd0) and a second CPMV difference value (mvd1) can be obtained based on a common difference predictor (mvdp) and a single absolute difference value (mvdd). can. For example, a first CPMV difference value (Mvd0) can be obtained based on the sum of mvdp and mvdd. Also, a first CPMV difference value (mvd1) can be obtained based on the sum of mvdp and -mvdd.

For example, an encoder can signal a common differential predictor (mvdp) and a differential absolute value (mvdd). A decoder can obtain a CPMV difference value corresponding to each of a plurality of control points of the current block based on a common difference predictor (mvdp) and a difference absolute value (mvdd). Although FIG. 52 is shown for 4-parameter affine motion compensation, this disclosure is not so limited. For example, when 6-parameter affine motion compensation is performed, some of the CPMVs included in the CPMV set of the current block can be derived using CPMV predictors of other CPMVs and CPMV difference values.

According to an embodiment of the present invention, the current block can be additionally partitioned according to the prediction mode of the current block. For example, if the prediction mode of the current block is the intra prediction mode, the encoder and decoder can divide the current block into a plurality of sub-blocks and perform intra prediction on each sub-block. A decoder can receive the aforementioned intra-prediction mode information from the bitstream. A decoder can determine whether to divide a current block into a plurality of sub-blocks based on intra-prediction mode information. At this time, information on how the current block is divided into a plurality of sub-blocks can be additionally signaled. Alternatively, the encoder and decoder may divide the current block using a preset method between the encoder and decoder according to the intra-prediction mode information. Next, the decoder can perform intra prediction for the current block or each of the plurality of sub-blocks based on the intra prediction mode information of the current block. Hereinafter, in the embodiments of FIGS. 53 to 60, the current block may be used as a term indicating a coding unit. Also, SCU (Sub-CU) in FIGS. 53 to 60 indicates a plurality of sub-blocks divided from the current block. Hereinafter, a detailed description will be given of how the current block is divided into a plurality of sub-blocks.

FIG. 53 illustrates various embodiments in which the current block is divided into a plurality of sub-blocks. FIGS. 53(a) and 53(b) illustrate sub-blocks divided into squares from the current block. For example, the current block can be divided into sub-blocks of preset size. At this time, the preset size may be a size pre-negotiated between the encoder and the decoder. According to one embodiment, the preset size may be an N×N size having the same height and width. For example, the preset size can be 4×4. According to one embodiment, the preset size may be a value set based on the size of the transform kernel. The preset size may be a value set based on the transform unit. The preset size can be the size of the transform unit.

According to one embodiment, the current block may be divided into sub-blocks of sizes determined by the size of the current block. For example, if the size of the current block is larger than a first threshold, the current block can be divided into sub-blocks of larger size than blocks smaller than the first threshold. Also, the operation of dividing into sub-blocks may be restricted according to the size of the current block. For example, if the size of the current block is less than a second threshold, the encoder and decoder may not split the current block into multiple blocks. This is because when the size of the current block is relatively small, the performance gain according to the method of performing intra prediction by dividing the current block may not be as large as when the size of the current block is relatively large.

According to one embodiment, the form in which the sub-blocks are divided can be determined independently of the form of the current block. For example, even if the current block is square, the shape of a plurality of sub-blocks divided from the current block may be non-square. Figures 53(c) and 53(d) illustrate non-square partitioned sub-blocks from the current block. According to one embodiment, the current block can be divided into non-square sub-blocks of preset size. At this time, the height and width of the preset size may be different from each other. For example, the preset size can be 4x8 or 8x4. As described above, the preset size may be a value set according to the size of the current block. According to one embodiment, the encoder and decoder can support non-square transform kernels of 4x8 or 8x4 size. Also, if the current block needs to be divided into non-square sub-blocks, the preset size may be a value set based on the non-square transform unit.

According to another embodiment, the current block can be divided into a plurality of sub-blocks having a form similar to that of the current block. Figures 53(e) and 53(f) illustrate sub-blocks divided from the current block when the current block is a non-square block. For example, if the current block is non-square, the current block can be divided into non-square sub-blocks similar in shape to the current block. As shown in FIG. 53(e), if the height of the current block is longer than the width, the current block can be divided into a plurality of sub-blocks with a height longer than the width. As shown in FIG. 53(f), when the width of the current block is longer than the height, the current block can be divided into a plurality of sub-blocks having a width longer than the height.

According to an embodiment of the present invention, the form in which the current block is divided can be determined based on the prediction mode of the current block. For example, at least one of the size or direction in which the current block is divided may vary depending on the intra-prediction mode (angle mode, DC mode, or planar mode) of the current block. FIG. 54 illustrates how the current block is divided according to the intra prediction mode of the current block. Referring to FIG. 54(a), a plurality of intra prediction modes can be classified into non-directional modes such as planar mode or DC mode and angular modes corresponding to intra prediction mode indexes 2 to 66, respectively.

According to an embodiment, when a current block is intra-predicted, among sub-blocks divided from the current block, residual signals of sub-blocks far from reference samples of the current block may be relatively large. Accordingly, the reference samples of the current block can be divided so that the samples having similar distances belong to the same sub-block.

According to one embodiment, if the intra-prediction mode of the current block is the non-directional mode, the current block can be divided into a plurality of square sub-blocks as shown in FIG. 54(b). Also, if the intra prediction mode of the current block is any one of the horizontal diagonal mode, the diagonal mode, and the vertical diagonal mode, the current block can be divided into a plurality of square sub-blocks. In addition, if the intra prediction mode of the current block is one of the horizontal diagonal mode, the diagonal mode, and the vertical diagonal mode, the current block can be divided into a plurality of square sub-blocks. .

On the other hand, if the intra-prediction mode of the current block is horizontal mode or vertical mode, the current block is divided into a plurality of non-square sub-blocks as shown in FIG. 54(c) or 54(d) according to the intra-prediction mode of the current block. can. Also, if the intra prediction mode of the current block is one of the horizontal mode and the vertical mode, the current block can be divided into a plurality of non-square sub-blocks. Through this, the current block can be divided from the reference samples of the current block so that samples with similar distances belong to the same sub-block.

For example, if the prediction mode of the current block is either the horizontal mode or the angular mode around the horizontal mode, the current block can be divided as shown in FIG. 54(c). If the prediction mode of the current block is either the vertical mode or the angular mode around the vertical mode, the current block can be divided as shown in FIG. 54(d). Different prediction methods can be applied to sub-blocks far from the reference samples of the current block.

FIG. 55 illustrates how a current block is divided into a plurality of sub-blocks based on sample values of reference samples of the current block according to one embodiment of the present invention. According to one embodiment of the present invention, the current block can be partitioned based on the sample values of the reference samples of the current block. According to one embodiment, the encoder and decoder can determine reference sample edges based on the sample values of the reference samples of the current block. For example, the reference sample edge may be a reference point dividing a region where the sample value of the reference sample changes by more than a threshold. Also, the reference sample edge may indicate a point where the sample value difference between adjacent reference samples is greater than or equal to a threshold.

Referring to FIG. 55(a), the encoder and decoder can determine the reference sample edge of the current block by comparing the sample values of the reference samples of the current block. For example, a position of a reference sample having a difference from an average value of a preset number of reference samples around a specific reference sample that is greater than or equal to a threshold may be determined as a reference sample edge. Also, the encoder and decoder can determine one or more reference sample edges for each of the upper reference sample set and the left reference sample set. Here, the upper reference sample set of the current block may be a set including reference samples located in the upper line of the current block. Also, the left reference sample set of the current block may be a set including reference samples positioned on the left line of the current block.

Next, the encoder and decoder can divide the current block based on a line segment connecting the reference sample edges of the upper reference sample set and the reference sample edges of the left reference sample set. For example, when one reference sample edge is detected for each of the upper reference sample set and the left reference sample set, the current block can be divided into a total of two sub-blocks (SCU1 and SCU2). Also, when two or more reference sample edges are detected for each of the upper reference sample set and the left reference sample set, the current block can be divided into a plurality of sub-blocks.

Meanwhile, as described above, reference samples for intra prediction of the current block may include samples on a plurality of reference lines. Referring to FIGS. 55(b) and 55(c), a current block can be divided based on sample values of reference samples on a plurality of reference lines. For example, a reference sample edge of the current block can be determined based on sample values of reference samples on a plurality of reference lines. When reference samples on a plurality of reference lines are used, the shape of an image object composed of peripheral blocks can be grasped.

According to one embodiment of the present invention, the encoder and decoder can determine one or more reference sample edges per reference line. Referring to FIG. 55(b), the encoder and decoder can first divide the current block by extending a line segment connecting the reference sample edges of each upper reference line. Two sub-blocks (SCU1, SCU2) can be obtained by the first division. Also, the encoder and decoder may secondly divide the current block by extending a line segment connecting the reference sample edges of each left reference line. Two sub-blocks (SCU2, SCU3) can be obtained by the second division. Accordingly, the current block can be divided even if one of the reference sample edge for each upper reference line and the reference sample edge for each left reference line is not detected.

FIG. 55(c) shows that the current block is divided when either one of the reference sample edges for each upper reference line or the reference sample edges for each left reference line is not detected or the reference sample edges are not arranged in a straight line. show how it is done. As with the left reference lines in FIG. 55(c), reference sample edges may not be detected in some of the left reference lines. If reference sample edges are detected in at least two reference lines, the encoder and decoder can divide the current block based on the corresponding reference sample edges. In addition, the above-described embodiments can also be applied when reference sample edges are not detected in some reference lines among the upper reference lines.

Also, like the upper reference line in FIG. 55(c), the reference sample edges for each upper reference line may not be arranged in one straight line. In this case, the current block can be divided based on the intermediate values of the reference sample edges on the lines not adjacent to the current block and the line segments passing the reference sample edges on the lines adjacent to the current block. Although FIGS. 55(b) and 55(c) illustrate embodiments in which three reference lines are used, the disclosure is not so limited.

According to an embodiment of the present invention, a plurality of sub-blocks divided from the current block are later predicted as primary sub-blocks (PSBs) that are preferentially predicted according to the intra-prediction mode of the current block. can be classified into secondary sub-blocks (SSBs). Also, a sub-block group made up of PSBs is called a primary sub-block group (hereinafter referred to as a PSB group), and a sub-block group made up of SSBs is called a secondary sub-block group (hereinafter referred to as an SSB group). be able to.

FIG. 56 illustrates how primary sub-block groups and secondary sub-block groups are determined according to one embodiment of the present invention. According to an embodiment, the PSB group may be composed of sub-blocks within a preset distance from reference samples referred to for intra-prediction of the current block among a plurality of sub-blocks divided from the current block. Also, the SSB group may be composed of sub-blocks outside a preset distance from reference samples referred to for intra prediction of the current block among the plurality of sub-blocks.

Referring to FIG. 56, the current block can be divided into a total of 16 sub-blocks. If the intra prediction mode of the current block is any one of a horizontal diagonal mode, a vertical diagonal mode, and a peripheral angle mode between the horizontal diagonal mode and the vertical diagonal mode, the PSB group and the SSB of the current block Groups can be determined as shown in FIG. 56(a). That is, among the 16 sub-blocks divided from the current block, the upper right 4 sub-blocks can be SSB, and the remaining sub-blocks can be PSB.

Each of FIGS. 56(a) to 56(e) shows that when the intra prediction mode of the current block is any one of HDIA, HOR, DIA, VER, VDIA, and each peripheral angular mode, Figure 3 illustrates how the PSB and SSB groups are determined; PSB and SSB can be distinguished by considering relative distances from reference samples used in each prediction mode. However, the disclosure is not so limited and other divisions similar to the example described in FIG. 56 are possible.

According to one embodiment, the encoder and decoder may perform intra prediction first on the PSB based on the intra prediction mode of the current block. The encoder and decoder can then perform intra prediction for SSB. Various intra prediction methods for the SSB of the current block will now be described in detail.

According to one embodiment, the encoder and decoder can perform prediction for the SSB of the current block based on the reconstructed sample values of the PSB of the current block. For example, the encoder and decoder can first recover the PSB of the current block based on the intra-prediction mode and reference samples of the current block. Next, the encoder and decoder can perform intra-prediction for the SSB of the current block by referring to the previously reconstructed samples adjacent to each of the SSBs of the current block. The encoder and decoder can perform intra prediction for SSBs of the current block based on the intra prediction mode of the current block. At this time, the pre-restored samples may be samples included in each PSB. The samples included in each of the PSBs may be samples closer to the SSB of the current block than the reference samples of the current block. Accordingly, even when the same prediction mode is used for SSBs of the current block, prediction accuracy for SSBs can be improved compared to the existing method.

According to one embodiment, an encoder and a decoder can perform prediction using different intra prediction modes for each of the PSB group and the SSB group of the current block. For example, the PSB of the current block can be predicted based on the primary mode (PM), which is the intra prediction mode of the current block. The SSB of the current block can be predicted based on a variable secondary mode (SM). For example, SM may be an intra-prediction mode applied only to SSBs among a plurality of sub-blocks of a current block. At this time, the SM can signal separately. According to one embodiment, the SM can signal an offset form relative to the PM. For example, SM may be an intra prediction mode corresponding to an intra prediction mode index added with an offset based on PM. At this time, the offset may be a value within a preset maximum value. Through this, the amount of calculation for the encoder to select a specific offset can be reduced.

Also, if the PM of the current block is an angular mode and there is an opposite angular mode of the PM, the SSB of the current block can be predicted based on the opposite angular mode of the PM. For example, if PM is a horizontal diagonal mode with an intra prediction mode index of 2, SM may be a vertical diagonal mode with an intra prediction mode index of 66 . Conversely, if PM is a vertical diagonal mode with an intra prediction mode index of 66, SM may be a horizontal diagonal mode corresponding to an intra prediction mode index of 2.

According to another embodiment, SM may be the preset prediction mode. For example, the SSB of the current block can be predicted based on a preset prediction mode. As described above, the SSBs of the current block can be sub-blocks outside a preset distance from the reference samples of the current block. Accordingly, it may be beneficial in terms of residual signal to predict the SSB of the current block based on the non-directional mode. That is, the preconfigured SM can be either a planar mode or a DC mode.

According to an additional embodiment, the final predicted block for the SSB of the current block is based on the first predicted SSB predicted based on the PM of the current block and the second predicted SSB predicted based on the non-directional mode described above. can be obtained by For example, the encoder and decoder can obtain the final predicted block corresponding to the SSB based on the average between the first predicted SSB and the second predicted SSB of the current block. Also, the encoder and decoder can restore the SSB by summing the final prediction block corresponding to the SSB and the residual signal.

FIG. 57 illustrates how a primary group of sub-blocks and a secondary group of sub-blocks are predicted in accordance with one embodiment of the present invention. Referring to FIG. 57, the current block can be divided into a total of 4 sub-blocks. According to an embodiment, when the prediction mode of the current block is one of the diagonal mode and the angular mode around the diagonal mode, the PSB group and the SSB group of the current block are as shown in FIG. can decide. That is, among four subblocks divided from the current block, the lower right subblock (SCU4) is the SSB of the current block, and the remaining subblocks (SCU1, SCU2, SCU3) are the current block. can be a PSB of

In this case, the SSB of the current block can be restored by referring to the samples included in each of the PSBs of the current block. For example, the encoder and decoder can first recover the PSB of the current block based on the intra-prediction mode and reference samples of the current block. Next, the encoder and decoder can perform intra-prediction for the SSB using the sample values of the samples adjacent to the SSB among the previously reconstructed samples of the PSB of the current block. For example, the SSB (SCU4) of the current block can be reconstructed based on the sample values of the previously reconstructed samples of the PSBs (SCU1, SCU2, and SCU3) and the residual signal.

According to one embodiment, when the prediction mode of the current block is one of the horizontal diagonal mode and the angular mode around the horizontal diagonal mode, the PSB group and the SSB group of the current block are shown in FIG. can be determined as That is, among four subblocks divided from the current block, the upper right subblock (SCU2) is the SSB of the current block, and the remaining subblocks (SCU1, SCU3, SCU4) are the SSB of the current block. It can be a PSB. In this case, as described above, the SSB (SCU2) of the current block is based on the sample values of the samples adjacent to the SSB (SCU2) among the already-restored samples of the PSBs (SCU1, SCU3, and SCU4) of the current block. can be restored.

According to additional embodiments, some of the reference samples of the current block may be adjacent to the SSB of the current block. At this time, some of the reference samples of the current block may be reference samples that are not used for intra prediction of the current block. In this case, the SSB of the current block can be predicted based on SM, which is a prediction mode different from PM of the current block. For example, if the PM of the current block is angular mode, the SM of the current block can be the opposite angular mode of PM. As shown in FIG. 57(b), when the PM of the current block is in the horizontal diagonal mode, the SM can be in the vertical diagonal mode.

Also, if the PM of the current block is in angular mode, the final predicted block of the SSB of the current block is the first predicted SSB predicted based on the PM of the current block and the second predicted SSB predicted based on the opposite angular mode of the PM. 2 can be obtained based on the predicted SSB. It is possible to use the angular mode opposite to the PM mode as in FIGS. 56(a) and (e), and if there is a reference sample adjacent to the boundary of the SSB, the previously described embodiments are applicable.

As mentioned above, in the process of coding a video signal, pictures can be divided into a sequence of coding tree units (CTUs). The following describes the order in which CTUs are processed within a picture, slice, or tile. FIG. 58 illustrates the order in which coding tree units are processed according to one embodiment of the invention.

An encoder and a decoder can encode or decode CTUs in a current picture including the current block or slices/tiles divided from the current picture. At this time, the encoder and decoder can encode or decode the plurality of CTUs according to a predefined processing order. If the CTUs are processed in an order different from the predefined processing order, the encoder can signal the order to the decoder.

FIG. 58(a) shows a raster scan order in which horizontal processing is performed from the topmost left CTU of a picture, slice, or tile, and then the next row of CTUs is processed. FIG. 58(b) shows an embodiment in which the location of the first CTU (Block A) to be processed first in a picture, slice or tile is explicitly signaled. For example, the encoder can signal the location of the first CTU. The decoder can process the corresponding CTU with the highest priority based on the signaled position of the first CTU. A decoder can decode a CTU corresponding to a previous position in the raster scan order based on the position of the first CTU according to the inverse-raster scan order. In addition, CTUs corresponding to subsequent positions in the raster scan order based on the position of the first CTU can be decoded according to the raster scan order described above. At this time, the CTUs processed according to the reverse-raster scan order and the CTUs processed according to the raster scan order are each processed according to the intersecting order, or which one of the two CTUs has priority. can be processed effectively.

Figures 58(c) and 58(d) show embodiments in which the locations of preferentially processed CTUs in a picture, slice or tile are explicitly signaled. For example, the plurality of CTUs can include a second CTU (Block B) and a third CTU (Block C). The encoder can signal the position of each of the second CTU (Block B) and the third CTU (Block C). Also, the position of the second CTU to be processed first among the plurality of preferentially processed CTUs can be additionally signaled.

Referring to FIG. 58(d), the decoder can preferentially decode the second CTU (Block B) based on the signaled location of the second CTU. The decoder can then decode the third CTU (Block C). The CTUs between the second CTU location and the third CTU location can then be decoded. For example, the decoder can decode the CTUs from the next CTU of the second CTU to the previous CTU of the third CTU in raster scan order or reverse raster scan order. In addition, the decoder can decode the CTU corresponding to the previous position in the raster scan order according to the inverse-raster scan order based on the position of the second CTU (Block B). Also, based on the position of the third CTU (Block C), CTUs corresponding to subsequent positions in the raster scan order can be decoded according to the above-described raster scan order.

FIG. 58(c) shows yet another embodiment in which the locations of preferentially processed CTUs in a picture, slice or tile are explicitly signaled in a manner similar to FIG. 58(d).

FIG. 59 illustrates a bi-directional intra prediction method according to one embodiment of the present invention. According to one embodiment of the present invention, a current block can be predicted using a bi-directional intra prediction method according to the distribution of available reference samples of the current block. When the current block or a plurality of sub-blocks divided from the current block are encoded or decoded as shown in FIGS. The lower and right reconstructed samples can be used as reference samples for the corresponding block. Accordingly, a specific block can be predicted by a bidirectional intra prediction method using already reconstructed samples on the right and bottom sides of the block. Hereinafter, a specific block indicates a current block or a plurality of sub-blocks divided from the current block.

FIGS. 59(a) and 59(b) illustrate an embodiment of a method for predicting a current block when there are already reconstructed samples in all four sides of a specific block. Referring to FIG. 59(a), a specific block can be divided into two equal parts based on the prediction mode of the current block. For example, if the prediction mode of the current block is the angular mode, the specific block can be bisected based on a line segment perpendicular to the prediction mode of the current block. Also, the first part of the bisected specific block can be reconstructed based on the prediction mode of the current block and the first reference samples. At this time, the first part may be a part including the upper side and the left side of the specific block. Also, the first reference samples may be reference samples of left and upper lines of a specific block. A second portion of the bisected specific block can be reconstructed based on the secondary prediction mode and the second reference samples. At this time, the second part may be a part including the right side and the bottom side of the specific block. Also, the second reference samples may be reference samples of the right and bottom lines of a particular block. According to one embodiment, the secondary prediction mode may be SM as described with reference to FIGS. 56 and 57. FIG.

Referring to FIG. 59(b), a specific block may not be additionally divided. For example, a particular block can be reconstructed based on a first predicted block and a second predicted block. At this time, the first prediction block may be a prediction block predicted based on the prediction mode of the current block and the first reference samples. Also, the second prediction block may be a prediction block predicted based on the secondary prediction mode and the first reference samples. A decoder may reconstruct a specific block by weighting the first prediction block and the second prediction block.

FIGS. 59(c) and 59(d) illustrate an embodiment of a method of predicting a current block when there are already reconstructed samples in only three of four sides of a specific block. is. In this case, the specific block can be divided into two halves based on the prediction mode of the current block. For example, a region of a specific block corresponding to a side where there are no previously restored samples can be predicted by extending the prediction mode from other sides. Also, in the embodiments of FIGS. 59(c) and 59(d), as shown in FIG. 59(b), a specific block is divided based on a plurality of prediction blocks for the specific block without additionally dividing the specific block. Embodiments that restore are applicable.

FIG. 60 illustrates how each of a plurality of sub-blocks divided from a current block is predicted according to one embodiment of the present invention. According to an embodiment of the present invention, each of a plurality of sub-blocks divided from a current block can be predicted based on an intra prediction mode used for prediction of neighboring blocks of the current block.

FIG. 60(a) illustrates the positions (AL, A, AR, R, L, BL) of each of a plurality of neighboring blocks of the current block. At this time, there may be no intra-prediction mode corresponding to each position of the peripheral blocks. For example, the block at the corresponding position may be inter-predicted, or the block at the corresponding position may not be decoded according to the raster scan order or the CU processing order.

FIG. 60(b) shows how each sub-block is predicted when the current block is divided into a total of four sub-blocks. For example, the first sub-block SCU1 can be predicted using the intra prediction mode corresponding to the upper left position (AL) closest to the first sub-block SCU1. Also, the second sub-block SCU2 can be predicted using the intra prediction mode corresponding to the upper position (A) adjacent to the second sub-block SCU2. Also, if there is no intra prediction mode corresponding to the upper position (A), the second sub-block SCU2 can be predicted using intra prediction modes corresponding to other peripheral positions (eg, upper right position). Also, like the fourth sub-block SCU4, if both the upper side and the left side are within the current block, the corresponding block can be predicted using the intra-prediction mode of neighboring blocks located on a distant straight line.

FIG. 60(c) shows yet another method of predicting each sub-block when the current block is divided into a total of four sub-blocks. In FIG. 60(c), the first sub-block SCU1, the second sub-block SCU2 and the third sub-block SCU3 can be predicted based on the intra prediction mode of the current block. At this time, the first sub-block SCU1, the second sub-block SCU2, and the third sub-block SCU3 may be the primary sub-blocks described above with reference to FIGS. According to one embodiment, the fourth sub-block SCU4 can be predicted based on an intra-prediction mode different from the intra-prediction mode of the current block. For example, the fourth sub-block SCU4 can be predicted using the intra prediction mode corresponding to the right position (R). Alternatively, the fourth sub-block SCU4 can be predicted using the intra-prediction modes of neighboring blocks located in a distant line.

The embodiments of the present invention described above can be implemented through various means. For example, embodiments of the invention may be implemented in hardware, firmware, software, or a combination thereof.

In the case of a hardware implementation, methods according to embodiments of the present invention may be implemented using one or more of ASICs (Application Specific Integrated Circuits), DSPs (Digital Signal Processors), DSDPs (Digital Signal Processing Devices), PDLs (Programmable Dynamic Circuits). ), FPGAs (Field Programmable Gate Arrays), processors, controllers, microcontrollers, microprocessors, and the like.

In the case of a firmware or software implementation, the methods according to embodiments of the invention may be implemented in the form of modules, procedures or functions that perform the functions or operations described above. Software codes are stored in memory and implemented by a processor. The memory is located inside or outside the processor and exchanges data with the processor by various well-known means.

Some embodiments may also be embodied in the form of a recording medium containing computer-executable instructions, such as program modules, executed by a computer. Computer-readable media can be any available media that can be accessed by the computer and includes both volatile and nonvolatile media, removable and non-removable media. Also, computer-readable media includes both storage media and communication media. Computer storage media include volatile and non-volatile media embodied in any method or technology for storage of information such as computer-readable instructions, data structures, program modules, or other data, removable and non-separable media. Communication media typically embodies computer readable instructions, data structures, or other data in a modulated data signal, such as program modules, or other transmission mechanisms and includes any information delivery media.

The above description of the present invention is for illustrative purposes only, and those skilled in the art to which the present invention pertains may make other specific modifications without changing the technical spirit or essential features of the present invention. It should be understood that it can be easily modified to any form. Accordingly, the above-described embodiments are to be considered in all respects as illustrative and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and components described as being distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims below rather than the detailed description given above, and all changes or modifications derived from the meaning and scope of the claims and their equivalents are considered to be the present invention. should be construed as being within the scope of

110 converter
115 Quantizer
120 Inverse quantizer
125 Inverse Transformer
130 filtering section
150 predictor
152 Intra Predictor
154 Inter predictor
154a motion estimator
154b motion compensator
160 Entropy coding part
210 Entropy decoding unit
220 Inverse quantizer
225 Inverse Transformer
230 filtering section
250 predictor
252 Intra Predictor
254 Inter predictor
710 current picture
720 reference picture
810 Encoder
820 decoder

Claims (17)

In a video signal decoding device,
including a processor;
The processor
constructing a motion vector prediction (MVP) candidate list for motion compensation of the current block;
obtaining a motion vector predictor of the current block based on the constructed MVP candidate list;
obtaining a motion vector difference value indicating the difference between the motion vector of the current block and the motion vector predictor;
obtaining a first value of an indicator that indicates at least one of a first resolution of the motion vector difference values of the current block and a second resolution of the motion vector difference values of the current block;
the first resolution is one of a plurality of available resolutions included in a first resolution set, the second resolution is one of a plurality of available resolutions included in a second resolution set;
determining whether to use the first resolution set or the second resolution set based on a prediction mode of the current block;
both the first resolution set and the second resolution set include a resolution of 1 sample unit;
the sample-by-sample resolution included in the first resolution set and the sample-by-sample resolution included in the second resolution set are indicated by different values of the indicator;
modifying the motion vector difference value based on the resolution of the motion vector difference value of the current block;
the resolution of the motion vector difference value is the first resolution or the second resolution ;
obtaining a motion vector of the current block based on the motion vector predictor and the modified motion vector difference value;
A video signal decoding apparatus for reconstructing the current block based on the acquired motion vector. 2. The video signal decoding apparatus of claim 1, wherein the second resolution set includes at least one other available resolution than the plurality of available resolutions included in the first resolution set. if the prediction mode of the current block is based on an affine model, the resolution of the motion vector difference value is obtained from the first resolution set;
3. The video signal decoding apparatus of claim 2, wherein the resolution of the motion vector difference values is obtained from the second resolution set if the prediction mode of the current block is not based on the affine model. 4. The method of claim 3, wherein the largest first available resolution among the plurality of available resolutions included in the first resolution set is smaller than the largest second available resolution among the plurality of available resolutions included in the second resolution set. video signal decoding device. the largest first available resolution is a resolution in units of one sample;
5. The video signal decoding apparatus of claim 4, wherein the second largest available resolution is a resolution of 4 samples. if the prediction mode of the current block is based on an affine model , the first value indicates a first available resolution that is one of available resolutions included in a first resolution set;
if the prediction mode of the current block is not based on the affine model, the first value indicates a second available resolution that is one of the available resolutions included in a second resolution set;
2. The video signal decoding apparatus of claim 1 , wherein the first available resolution and the second available resolution are different from each other. The indicator is represented by bits of variable length,
2. The video signal decoding apparatus of claim 1 , wherein the first value is one of a plurality of values represented by the variable length bits. 3. The video signal decoding apparatus of claim 2, wherein the number of available resolutions included in the first resolution set and the number of available resolutions included in the second resolution set are different from each other. The number of available resolutions included in the first resolution set and the number of available resolutions included in the second resolution set are determined based on a picture order count (POC) of reference pictures. 9. Video signal decoding apparatus according to claim 8 . obtaining a resolution of a motion vector difference value of the current block from the first resolution set if the picture order count of the reference picture is the same as the POC of the current picture containing the current block;
if the picture order count of the reference picture is not the same as the POC of the current picture containing the current block, obtain the resolution of motion vector difference values of the current block from the second resolution set;
10. The video signal of claim 9 , wherein the first resolution set includes available resolutions excluding the smallest available resolution and the second smallest available resolution among the available resolutions included in the second resolution set. decoding device. In a video signal encoding device,
including a processor;
The processor
obtaining a bitstream decoded by a decoder using a decoding method;
The decryption method includes:
constructing a motion vector prediction (MVP) candidate list for motion compensation of the current block;
obtaining a motion vector predictor for the current block based on the constructed MVP candidate list;
obtain a motion vector difference value that indicates the difference between the motion vector of the current block and the motion vector predictor ;
obtaining a first value of an indicator that indicates at least one of a first resolution of the motion vector difference values of the current block and a second resolution of the motion vector difference values of the current block;
the first resolution is one of a plurality of available resolutions included in a first resolution set, the second resolution is one of a plurality of available resolutions included in a second resolution set;
determining whether to use the first resolution set or the second resolution set based on a prediction mode of the current block;
both the first resolution set and the second resolution set include a resolution of 1 sample unit;
the sample-by-sample resolution included in the first resolution set and the sample-by-sample resolution included in the second resolution set are indicated by different values of the indicator;
modifying the motion vector difference value based on the resolution of the motion vector difference value of the current block;
the resolution of the motion vector difference value is the first resolution or the second resolution;
obtaining a motion vector of the current block based on the motion vector predictor and the modified motion vector difference value;
A video signal encoding apparatus for reconstructing the current block based on the obtained motion vector . 12. The video signal encoding apparatus of claim 11 , wherein the second resolution set includes at least one other available resolution than the plurality of available resolutions included in the first resolution set. 13. The method of claim 12 , wherein the largest first available resolution among the plurality of available resolutions included in the first resolution set is smaller than the largest second available resolution among the plurality of available resolutions included in the second resolution set. video signal encoding equipment. 14. The video signal encoding apparatus of claim 13 , wherein the largest first available resolution is a resolution in units of 1 sample and the second largest available resolution is a resolution in units of 4 samples . The indicator is represented by bits of variable length,
12. The video signal encoding apparatus of claim 11 , wherein the first value is one of a plurality of values represented by the variable length bits. 13. The video signal encoding apparatus of claim 12 , wherein the number of available resolutions included in the first resolution set and the number of available resolutions included in the second resolution set are different from each other . A method for obtaining a bitstream comprising:
constructing a motion vector prediction (MVP) candidate list for motion compensation of the current block;
obtaining a motion vector predictor for the current block based on the constructed MVP candidate list;
obtaining a motion vector difference value indicating the difference between the motion vector of the current block and the motion vector predictor;
obtaining a first value of an indicator that indicates at least one of a first resolution of the motion vector difference values of the current block and a second resolution of the motion vector difference values of the current block; The first resolution is one of a plurality of available resolutions included in a first resolution set, the second resolution is one of a plurality of available resolutions included in a second resolution set, and the first resolution whether to use a set or the second resolution set is determined based on the prediction mode of the current block, both the first resolution set and the second resolution set include a resolution in units of one sample; wherein the one-sample resolution included in the first resolution set and the one-sample resolution included in the second resolution set are indicated by different values of the indicator;
modifying the motion vector difference value based on the resolution of the motion vector difference value of the current block, wherein the resolution of the motion vector difference value is the first resolution or the second resolution; ,
obtaining a motion vector for the current block based on the motion vector predictor and the modified motion vector difference value;
obtaining a bitstream by decoding information for the motion vector of the current block .

Images